@heater: how does a neuron know in which way to grow its synapses?
Behold the holy grail of brain modeling. Firing is pretty well understood and it's not all that complicated, but the algorithms describing dendrite growth and synapse formation are pretty much completely unknown.
The going theory, at least among people who are not trying to use Parallax hardware to model it, is that there is a reasonable number of families of dendrite and synapse types and that these are guided by (1) areal attractant factors that guide growth to the right area of the brain or the right part of a "map" within an area; (2) likeness attractants that encourage synapse formation of either excitory or inhibitory synapses, either when neurons fire in tandem or in some cases when they don't. "Neurons that fire together, wire together," is a phrase you sometimes hear.
None of these individual algorithms is necessarily all that complex. It's the way they interact and the sheer number of neurons involved that make the result so interesting.
Leon: "It must be controlled by something inside the neuron"
I'm not sure what is controlling what. If there is a chemotactic thing going on then that "chemical soup" in the environment is exerting control over the neuron. Of course the neurons are contributing to the chemical soup...
Anyway the whole things seems more complicated than wiring up a bunch of simulated neurons in a neural net.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
as far as I know it the connections between neurons are selforganised.
Whenever a group of neurons receives similar patterns the connections between them get more intensive
depending on the USE of a synapse the contact-area of the synapse will grow bigger.
The second effekt of repeating is that the number of neurons belonging to a representing system will grow through repeating
A professional guitar-musician that has practised playing for more than 10.000 hours since he was a young child has a much bigger area inside his brain
that does represent his fingers in his brain than an average man or a hobby musician that has started practising as a teenager and has only 2000 hours of experience.
heater said...
Leon: "It must be controlled by something inside the neuron"
I'm not sure what is controlling what. If there is a chemotactic thing going on then that "chemical soup" in the environment is exerting control over the neuron. Of course the neurons are contributing to the chemical soup...
Anyway the whole things seems more complicated than wiring up a bunch of simulated neurons in a neural net.
I think it's a lot more likely to be something inside the cell than something in the surrounding environment. The latter will be fairly homogeneous.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
@Leon: The latter (surrounding environment) will be fairly homogeneous.
According to who?
It is known that chemical markers in the environment guide neuroblast migration during fetal development; some of the specific markers are even known. Some of the more horrific birth defects are due to toxins and genetic screw-ups that interfere with those markers.
For a cell on one side of your brain to grow a fibre which lands at a specific area on the other side of your brain (as many of the millions of miles of fibre in your brain manage to do well before you have any correlated activity to guide them) environmental markers are about the only plausible mechanism.
I believe the general consensus nowadays is that chemical markers guide long-distance fibre growth (that is, growth over distances of more than a tenth of a millimeter or so), while specific synapse formation and local landings are combination of random formation (a sort of initialization to get the real process started) and enhancement based on input correlations. But even the latter are mediated by local environmental chemical markers.
I don't know ANYBODY who would say the intracellular environment of the brain is in any meaningful sense "homogeneous."
It's more likely to be other cells, such as the glial cells, that direct dendritic growth. There are a lot more glial cells than neurons in the cerebral cortex. The glial cells can modulate the inter-cellular fluid on a limited, local level.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
@Leon: It's more likely to be other cells, such as the glial cells, that direct dendritic growth
While that may be true in some cases in extremely short range connectivity (and even that's very controversial), it buys you nothing when asking how a fibre from one side of the brain finds a landing area 20 cm away on the other side.
Given their lack of long-range connectivity I would be amazed to learn that the glia are doing anything more complex than amplifying marker densities originated by the neurons.
But glial cells communicate with each other, which could explain how long fibres find their way. They definitely play a large role during brain development.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
They modify the growth of axons and dendrites during development, and myelinate nerve fibres. Think of a sequence of glial cells, each communicating with its neighbours, this can be done over large distances. They carry on doing this in adulthood, when neural connections are made during learning. Unlike neurons, some glial cells can replicate themselves.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
@Leon: They (glial cells) do it (communicate with each other) during development, over several cm if necessary, by modifying the growth of axons and dendrites.
WHAT? I cannot see how you have done anything except exchange the goalposts here. If the original problem is "how does neuron A grow a dendrite to neuron B across a macroscopic divide," your answer is "the glial cells do it" and now that the problem is "how does glial cell A wave at glial cell B across a macroscopic divide so that it can attract the dendrite" your answer is "the neurons do it?"
...some glial cells can replicate themselves
This is irrelevant to anything, but it is probably worth mentioning that neurons can be replaced too; it was discovered a few years that the stem cells for this are part of the bone marrow.
on edit, Leon added...
...Think of a sequence of glial cells, each communicating with its neighbours, this can be done over large distances.
Okay, that clarifies things; you're proposing that they use TCP.· I find chemical attractants to be much more believable.
I read a report about a 3 year old child that was in danger of dying
from an infection in the brain. HALF of the brain was amputated and now
beeing seven years old she lives a almost normal life !
It isn't a matter of cell replacement. The developing brain is highly "plastic": when it is damaged other parts of the brain take over the functions that were performed by the damaged portions.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Yes it is. The researchers injected floursecent dye into the bone marrow of rats, and after dissecting the rats found that their brains glowed in the dark. Microscopic examination revealed that it was a subset of the neurons that were glowing. The conclusion is that new neurons grow from stem cells located in the bone marrow. That wasn't all, they realized it was a pretty extraordinary conclusion and did a bunch of other tests to verify that's what was happening.
This is a google books result, url might not work but whatever...
I still do not see how glial cell intercommunication could be meaningful over a distance of more than a few microns. Neurons at least can be shown to have *some* method of targeted long-distance communication.
The preponderance of glial cells is what makes it IM-plausible. There are over a hundred billion glial cells in the brain,
and they DON'T have any known means of targeted long-distance intercommunication. By contrast, chemical attractant
markers are known to exist and to be used, particularly during fetal development when things are still relatively simple
and amenable to our existing research methods. Saying "the glial cells guide the dendrites" is just hand-waving;
it really opens a much harder question than asking how the neurons do that without the glial cells.
You're suggesting that glial cells somehow hand off directional and target information cell after cell for literally
millions of iterations, and somehow manage to keep thousands of such messages separate from one another as
they pass in different directions. I do not know a single person seriously investigating brain function who would
not laugh out loud at this idea (and I've had quite a few more online conversations with those people than most
people ever do).
By contrast, it is KNOWN that chemical markers can act with extreme specificity and sensitivity, and a few of
these are known from their role in fetal development. Why you are so resistant to this idea, which is currently
the mainstream consensus, mystifies me.
Noticed later: Is there any evidence that large scale replacement of damaged brain tissue takes place in humans?
No, just as there isn't any evidence that a large chunk of your arm can be replaced if it is cut off. This has nothing to do with whether
individual cells can be replaced within an otherwise intact organ; both arms and brains appear to be able to do that. Curiously, amphibians
seem to be able to replace macroscopic amounts of both arm and brain that have been removed, which suggests that there is no
fundamental reason we couldn't be made to do both too, perhaps with gene therapy once we've learned the trick.
Comments
Behold the holy grail of brain modeling. Firing is pretty well understood and it's not all that complicated, but the algorithms describing dendrite growth and synapse formation are pretty much completely unknown.
The going theory, at least among people who are not trying to use Parallax hardware to model it, is that there is a reasonable number of families of dendrite and synapse types and that these are guided by (1) areal attractant factors that guide growth to the right area of the brain or the right part of a "map" within an area; (2) likeness attractants that encourage synapse formation of either excitory or inhibitory synapses, either when neurons fire in tandem or in some cases when they don't. "Neurons that fire together, wire together," is a phrase you sometimes hear.
None of these individual algorithms is necessarily all that complex. It's the way they interact and the sheer number of neurons involved that make the result so interesting.
I'm not sure what is controlling what. If there is a chemotactic thing going on then that "chemical soup" in the environment is exerting control over the neuron. Of course the neurons are contributing to the chemical soup...
Anyway the whole things seems more complicated than wiring up a bunch of simulated neurons in a neural net.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
I did say he was in bed. That guy was sleeping, he dreams in Morse code !!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Whenever a group of neurons receives similar patterns the connections between them get more intensive
depending on the USE of a synapse the contact-area of the synapse will grow bigger.
The second effekt of repeating is that the number of neurons belonging to a representing system will grow through repeating
A professional guitar-musician that has practised playing for more than 10.000 hours since he was a young child has a much bigger area inside his brain
that does represent his fingers in his brain than an average man or a hobby musician that has started practising as a teenager and has only 2000 hours of experience.
best regards
Stefan
I think it's a lot more likely to be something inside the cell than something in the surrounding environment. The latter will be fairly homogeneous.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
According to who?
It is known that chemical markers in the environment guide neuroblast migration during fetal development; some of the specific markers are even known. Some of the more horrific birth defects are due to toxins and genetic screw-ups that interfere with those markers.
For a cell on one side of your brain to grow a fibre which lands at a specific area on the other side of your brain (as many of the millions of miles of fibre in your brain manage to do well before you have any correlated activity to guide them) environmental markers are about the only plausible mechanism.
I believe the general consensus nowadays is that chemical markers guide long-distance fibre growth (that is, growth over distances of more than a tenth of a millimeter or so), while specific synapse formation and local landings are combination of random formation (a sort of initialization to get the real process started) and enhancement based on input correlations. But even the latter are mediated by local environmental chemical markers.
I don't know ANYBODY who would say the intracellular environment of the brain is in any meaningful sense "homogeneous."
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/24/2009 3:56:03 PM GMT
Damit, that's another thing we have to add to our neural net software.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
While that may be true in some cases in extremely short range connectivity (and even that's very controversial), it buys you nothing when asking how a fibre from one side of the brain finds a landing area 20 cm away on the other side.
Given their lack of long-range connectivity I would be amazed to learn that the glia are doing anything more complex than amplifying marker densities originated by the neurons.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/24/2009 5:03:27 PM GMT
So? Over what distance? Unless you are proposing that they use radio or TCP I really can't see how they do anything over macroscopic distances.
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/24/2009 6:07:16 PM GMT
WHAT? I cannot see how you have done anything except exchange the goalposts here. If the original problem is "how does neuron A grow a dendrite to neuron B across a macroscopic divide," your answer is "the glial cells do it" and now that the problem is "how does glial cell A wave at glial cell B across a macroscopic divide so that it can attract the dendrite" your answer is "the neurons do it?"
...some glial cells can replicate themselves
This is irrelevant to anything, but it is probably worth mentioning that neurons can be replaced too; it was discovered a few years that the stem cells for this are part of the bone marrow.
on edit, Leon added...
...Think of a sequence of glial cells, each communicating with its neighbours, this can be done over large distances.
Okay, that clarifies things; you're proposing that they use TCP.· I find chemical attractants to be much more believable.
college.usc.edu/news/stories/96/long-unappreciated-and-ignored-brains-glial-cells-get-their-due/
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
I read a report about a 3 year old child that was in danger of dying
from an infection in the brain. HALF of the brain was amputated and now
beeing seven years old she lives a almost normal life !
best regards
Stefan
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Yes it is. The researchers injected floursecent dye into the bone marrow of rats, and after dissecting the rats found that their brains glowed in the dark. Microscopic examination revealed that it was a subset of the neurons that were glowing. The conclusion is that new neurons grow from stem cells located in the bone marrow. That wasn't all, they realized it was a pretty extraordinary conclusion and did a bunch of other tests to verify that's what was happening.
This is a google books result, url might not work but whatever...
http://books.google.com/books?id=4-KHzWwgMKEC&pg=PA79&lpg=PA79&dq=bone+marrow+neural+stem+cells&source=bl&ots=bDdYUSPU5o&sig=96ROpk1Uu9A8w8P2TZr3tg8Vuzw&hl=en&ei=hPeSSrX1MIGANqbajZIK&sa=X&oi=book_result&ct=result&resnum=3#v=onepage&q=bone marrow neural stem cells&f=false
I still do not see how glial cell intercommunication could be meaningful over a distance of more than a few microns. Neurons at least can be shown to have *some* method of targeted long-distance communication.
Given the preponderance of glial cells, it seems plausible. Do you have an alternative explanation?
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
Post Edited (Leon) : 8/24/2009 8:47:20 PM GMT
and they DON'T have any known means of targeted long-distance intercommunication. By contrast, chemical attractant
markers are known to exist and to be used, particularly during fetal development when things are still relatively simple
and amenable to our existing research methods. Saying "the glial cells guide the dendrites" is just hand-waving;
it really opens a much harder question than asking how the neurons do that without the glial cells.
You're suggesting that glial cells somehow hand off directional and target information cell after cell for literally
millions of iterations, and somehow manage to keep thousands of such messages separate from one another as
they pass in different directions. I do not know a single person seriously investigating brain function who would
not laugh out loud at this idea (and I've had quite a few more online conversations with those people than most
people ever do).
By contrast, it is KNOWN that chemical markers can act with extreme specificity and sensitivity, and a few of
these are known from their role in fetal development. Why you are so resistant to this idea, which is currently
the mainstream consensus, mystifies me.
No, just as there isn't any evidence that a large chunk of your arm can be replaced if it is cut off. This has nothing to do with whether
individual cells can be replaced within an otherwise intact organ; both arms and brains appear to be able to do that. Curiously, amphibians
seem to be able to replace macroscopic amounts of both arm and brain that have been removed, which suggests that there is no
fundamental reason we couldn't be made to do both too, perhaps with gene therapy once we've learned the trick.