4 ms·

Glial brain cells, long in neurons’ shadow, reveal hidden powers

136 points by theafh 7y ago

DoctorOetker 7y ago
Could the glial cells be somehow responsible for backpropagation in the brain?
- nshm 7y ago
  Brain doesn't do backpropagation
  - chrisb 7y ago
    Interesting paper here on e-prop which is a biologically plausible alternative to backprop: https://www.biorxiv.org/content/10.1101/738385v3 https://www.biorxiv.org/content/10.1101/738385v3
  - DailyHN 7y ago
    > Brain doesn't do backpropagation What makes you so sure?
    - Sommer 7y ago
      If it did I would have learned from my mistakes.
      - short_sells_poo 7y ago
        He could have overfit due to his training size being too small. Either he has to add some noise or increase his set.
      - gbrown 7y ago
        Maybe your reward function is just wonky.
      - tudorw 7y ago
        That's a gene thing, apparently :) https://www.telegraph.co.uk/news/science/science-news/3317617/Failing-to-learn-from-mistakes-is-genetic.html https://www.telegraph.co.uk/news/science/science-news/331761...
  - cududa 7y ago
    You ever touch a hot stove/ iron/ oven as a kid? Do you still do that?
neural_thing 7y ago
Went to a conference a couple of years ago where there was an interesting talk on the subject. See here if you are interested: https://vimeo.com/262522684 https://vimeo.com/262522684
ickelbawd 7y ago
Does this mean the connectome is doomed or faulty since it maps only neurons and not glial cells?
- kingkawn 7y ago
  Not to mention the myriad other processes involved, known or not
- GliaApostle 7y ago
  I wouldn't say that. I'm on a team that's doing reconstruction: glia are included. Astrocytes have a massive surface area, so their meshes are harder to analyze, but we have them along with oligodendrocytes and some possible microglia. I would caution that the connectome is like the human genome. Both are very useful, but both came with a lot of hype that it would solve everything. We have connectomes for things like C. Elegans and we're still unsure about neuronal computations within them.
- cyorir 7y ago
  Not so much faulty as incomplete. There is still value in a partial picture.
- skohan 7y ago
  This is one of the reasons I have doubts when claims are made that we're anywhere close to AGI: the brain in vivo is incredibly parsimonious. Almost everything seems to be involved in information processing, down to sub-cellular machinery. I have a huge interest projects to accurately map and model the neuronal structure of the brain, but I suspect at some point they will reveal how much further fractal complexity there is to the human brain than we have ever imagined which we are yet unable to model.
jedberg 7y ago
In elementary school I had to do a report on Einstein, and I remember one of the important details was that he had a higher ratio of glial cells to neurons than a normal brain. At the time, they had no idea if this made any difference to his genius, and it sounds like they still aren't too sure.
Fishysoup 7y ago
We've known for at least a decade that astrocytes and other glia play important roles in synaptic transmission and learning, not just axon insulation and blood-brain-barrier related stuff. It's still unclear what most of their contributions are. They're harder to interrogate because they're not as electrically active as neurons, and though they do exhibit Calcium waves, they're small and seem to be localized to small compartments. There's a number of people out there working on glia in a functional sense, like Atwell at UCL and Magistretti (though he's retired now i think).
- api 7y ago
  We've known even longer than that: https://en.wikipedia.org/wiki/Tripartite_synapse https://en.wikipedia.org/wiki/Tripartite_synapse Earliest references are from the late 1990s.
  - Fishysoup 7y ago
    Oh wow didn't know it went back that early, nice
reasonattlm 7y ago
One of those hidden powers is the destruction of the aging brain. Senescence of glial cells appears to be quite important in the progression of neurodegeneration, as both a source of neuroinflammation and all of its woes, and a mechanism capable of more directly causing neuron death. Clearance of senescent glial cells prevents tau-dependent pathology and cognitive decline - https://doi.org/10.1038/s41586-018-0543-y https://doi.org/10.1038/s41586-018-0543-y Tau protein aggregation is associated with cellular senescence in the brain - https://doi.org/10.1111/acel.12840 https://doi.org/10.1111/acel.12840 Flushing out 'zombie cells' could help stave off Parkinson's, study suggests - https://www.theguardian.com/science/2018/jan/23/flushing-out-zombie-cells-could-help-stave-off-parkinsons-study-suggests https://www.theguardian.com/science/2018/jan/23/flushing-out... Astrocyte senescence: Evidence and significance - https://doi.org/10.1111/acel.12937 https://doi.org/10.1111/acel.12937 The good news here is that the senolytic drug dasatinib crosses the blood-brain barrier and has been demonstrated to destroy senescent glial cells in animal models, and destroy senescent cells elsewhere in the body in human trials. The cheap dasatinib / quercetin combination should be trialed in Alzheimer's and Parkinson's patients, and the sooner the better.
SeanFerree 7y ago
Very cool!
IIAOPSW 7y ago
I used to have fun trolling by saying "we only think with 10% of our brains". Then as soon as someone comes in thinking they were about to school me real good, I'd drop the fact that 90% of our brains aren't neurons but merely glial support structures so the statistic is technically true. The ol' double cross in pedantic trivia. I guess I need to stop doing that now.
- dwaltrip 7y ago
  I think the latest research indicates that glial cells are about as numerous as neurons (the article says something to this effect as well), so the joke would update to "we only use half of our brains".

136 points by theafh 7y ago

DoctorOetker 7y ago

Could the glial cells be somehow responsible for backpropagation in the brain?

nshm 7y ago
Brain doesn't do backpropagation
- chrisb 7y ago
  Interesting paper here on e-prop which is a biologically plausible alternative to backprop: https://www.biorxiv.org/content/10.1101/738385v3 https://www.biorxiv.org/content/10.1101/738385v3
- DailyHN 7y ago
  > Brain doesn't do backpropagation What makes you so sure?
  - Sommer 7y ago
    If it did I would have learned from my mistakes.
    - short_sells_poo 7y ago
      He could have overfit due to his training size being too small. Either he has to add some noise or increase his set.
    - gbrown 7y ago
      Maybe your reward function is just wonky.
    - tudorw 7y ago
      That's a gene thing, apparently :) https://www.telegraph.co.uk/news/science/science-news/3317617/Failing-to-learn-from-mistakes-is-genetic.html https://www.telegraph.co.uk/news/science/science-news/331761...
- cududa 7y ago
  You ever touch a hot stove/ iron/ oven as a kid? Do you still do that?

neural_thing 7y ago

Went to a conference a couple of years ago where there was an interesting talk on the subject. See here if you are interested: https://vimeo.com/262522684 https://vimeo.com/262522684

ickelbawd 7y ago

Does this mean the connectome is doomed or faulty since it maps only neurons and not glial cells?

kingkawn 7y ago
Not to mention the myriad other processes involved, known or not
GliaApostle 7y ago
I wouldn't say that. I'm on a team that's doing reconstruction: glia are included. Astrocytes have a massive surface area, so their meshes are harder to analyze, but we have them along with oligodendrocytes and some possible microglia. I would caution that the connectome is like the human genome. Both are very useful, but both came with a lot of hype that it would solve everything. We have connectomes for things like C. Elegans and we're still unsure about neuronal computations within them.
cyorir 7y ago
Not so much faulty as incomplete. There is still value in a partial picture.
skohan 7y ago
This is one of the reasons I have doubts when claims are made that we're anywhere close to AGI: the brain in vivo is incredibly parsimonious. Almost everything seems to be involved in information processing, down to sub-cellular machinery. I have a huge interest projects to accurately map and model the neuronal structure of the brain, but I suspect at some point they will reveal how much further fractal complexity there is to the human brain than we have ever imagined which we are yet unable to model.

jedberg 7y ago

In elementary school I had to do a report on Einstein, and I remember one of the important details was that he had a higher ratio of glial cells to neurons than a normal brain. At the time, they had no idea if this made any difference to his genius, and it sounds like they still aren't too sure.

Fishysoup 7y ago

We've known for at least a decade that astrocytes and other glia play important roles in synaptic transmission and learning, not just axon insulation and blood-brain-barrier related stuff. It's still unclear what most of their contributions are. They're harder to interrogate because they're not as electrically active as neurons, and though they do exhibit Calcium waves, they're small and seem to be localized to small compartments. There's a number of people out there working on glia in a functional sense, like Atwell at UCL and Magistretti (though he's retired now i think).

api 7y ago
We've known even longer than that: https://en.wikipedia.org/wiki/Tripartite_synapse https://en.wikipedia.org/wiki/Tripartite_synapse Earliest references are from the late 1990s.
- Fishysoup 7y ago
  Oh wow didn't know it went back that early, nice

reasonattlm 7y ago

One of those hidden powers is the destruction of the aging brain. Senescence of glial cells appears to be quite important in the progression of neurodegeneration, as both a source of neuroinflammation and all of its woes, and a mechanism capable of more directly causing neuron death. Clearance of senescent glial cells prevents tau-dependent pathology and cognitive decline - https://doi.org/10.1038/s41586-018-0543-y https://doi.org/10.1038/s41586-018-0543-y Tau protein aggregation is associated with cellular senescence in the brain - https://doi.org/10.1111/acel.12840 https://doi.org/10.1111/acel.12840 Flushing out 'zombie cells' could help stave off Parkinson's, study suggests - https://www.theguardian.com/science/2018/jan/23/flushing-out-zombie-cells-could-help-stave-off-parkinsons-study-suggests https://www.theguardian.com/science/2018/jan/23/flushing-out... Astrocyte senescence: Evidence and significance - https://doi.org/10.1111/acel.12937 https://doi.org/10.1111/acel.12937 The good news here is that the senolytic drug dasatinib crosses the blood-brain barrier and has been demonstrated to destroy senescent glial cells in animal models, and destroy senescent cells elsewhere in the body in human trials. The cheap dasatinib / quercetin combination should be trialed in Alzheimer's and Parkinson's patients, and the sooner the better.

SeanFerree 7y ago

Very cool!

IIAOPSW 7y ago

I used to have fun trolling by saying "we only think with 10% of our brains". Then as soon as someone comes in thinking they were about to school me real good, I'd drop the fact that 90% of our brains aren't neurons but merely glial support structures so the statistic is technically true. The ol' double cross in pedantic trivia. I guess I need to stop doing that now.

dwaltrip 7y ago
I think the latest research indicates that glial cells are about as numerous as neurons (the article says something to this effect as well), so the joke would update to "we only use half of our brains".