Neural Connections

[Mythics]

I'm just curious here, but has anyone actually discovered any of the following?

What causes neurons to connect to one another? (Some neurons manage to have their axon travel rather great distances to reach one specific neuron located half-way across the brain.)

If a neuron sits at it's resting potential for 'too long', does it or any part of it degrade/decay/move/change it's path?

Do neurons regularly rebuild paths between themselves? Do they disconnect and reconnect to others at all even?

I dabble in programming occasionally and have put together a few Artificial Neural Networks, but in each of them the structure of the paths are so very limited and only seem to be able to simulate one aspect of how the human brain actually functions. Considering so many aspects of neurons are similar if not identical throughout the brain, regardless of their use, it would be nice to know how exactly the structure of the network itself is formed rather than so much documentation being so concentrated on how neurons themselves work.

If, for example, a neuron's path through the network is achieved based on the firing of other neurons.. that would be a tremendous piece of knowledge for myself.

If a neuron is not activated, especially a set of physically close neurons, within some time frame and therefore attempts to move and/or disconnect itself to reconnect elsewhere..

Or if a nueron's initial connection is derived from it's terminal seeking out non-firing neurons and/or rapidly firing neurons.. that'd just be nifty to know.

(I may be stating the obvious here, but considering my lack of knowledge, I'm just seeking any form of discussion and/or answers I can find.)

Thanks in advance.

[phil123424]

Hallo,

the obvious problems might be the hard problems

In my opinion the neuronal network of the brain is based of the development of maps or models of neuronal circuits which are conserved in the genetic code. First of all the developing brain is a mass of high potential cells, which can devide into the specific neurons, regulated by growing factors (f.e. NGF), transcritions factsors (f.e.HOX.) etc.. These factors are, of course, regulated by the genetic code, the enviroment, time and the individual aspects (etc.). A factors which leads one neuron to the other is called Reelin, produced by specified neurons to attract the proper neuro to his new place and enviroment in the brain, and further on let the axon spread and connect to its new buddy-neuron.
Thats placed in the developing brain of the fetus. But the brain is developing itself after birth, too.

It`s propably not possible to change this models or circuits, but to connects circuits with other neurons or circuits to create a new integrity of functions. After birth many of the cells in the brain die from apoptosis to enable an more effective and individual brain function.
So the cause of the a new connecting is learning. But in my opinion it would be not very effective to create knew guidelines half-way across the brain (after birth), when instead you can use existing routs, or spare existing routs from apoptosis.

When a neuron sits at its resting potential for too long then it dies. The neurons which are not integrated in a active circuit or collum die, because first the synapses get weaker, and the the metabolic status get less, after a time the neuron dies (from apoptosis, or negrosis).

But the neuron has normally between 1000-20000 synaptical connectings, therefore it`s possible to loose connections and get reconnected to another circuit or columm( which in my opionins gives the function). By associative learning one neuron can be a part of associative fields. In the associative parts of our brain different brain areas are connected widely, if you have a widespread associative thinking you are more creative.

Anyway, the questions of you are quite difficult to sole even today. Some of them are not totally understood.

What do you try to find with programming. Which questions are you following? It`s hard for me to unterstand which sponatnious answers a computer can give us even if it able to create spontaneous action impulses. But still a computer is not a living thing, its hardware is not biological integrated like a cell is. Of course the neurophsics are interesting and can lead to a some interesting new hypothesises, but conciousness can not be unterstood when we look at the feedback of a calculating processor.
But ok, I don`t know much about that field. But tell me what do they try to find behind this artificial networks?

Kind regards

Phil

[Mythics]

Thank you for your reply Phil. While a bit of the terminology you used is unfamiliar to me now, I do intend to research these things further.

Regarding artificial neural networks in programming, they have a variety of purposes. Most of the time, they are only based on bits and pieces of what we know of biological neurons and their functions. Quite often, they are also manually created and adjusted to find answers as well.

There are many types of artificial neural networks out there that behave very differently to solve many different kinds of problems. One quite common use is handwriting recognition. I would strongly recommend checking out the back propagation method as it's quite easy to understand.

I don't personally have any kind of problems I'm trying to work out using them, I'm just interested in the workings of them and the mind as well as potentially creating my own type of Artificial Neural Network that may be closer to the way the human mind actually behaves. I've seen so many different types of Artificial Neural Networks that closely resemble basic aspects of the way many networks within the brain function, but rarely have I seen or heard of any that can adjust themselves to function differently based on the 'problems' involved.

[StriatumPDM]

What phil means is that there are gradients of chemical substances in nervous systems which can either attract or repel neurons or their axons. Neurons possess so-called growth cones that may sense these substances with special receptors. The binding of the substance may then lead to a reconstructuring of the neuron's cytoskeleton so that the growth cone either moves towards the higher gradient of the substance or moves away from it. I attached an image which illustrates this mechanism.

Reelin, the substance phil mentioned, is such an important chemical important for the development of the cortex. "Newly-born" neurons grow (or migrate) from the inside of the brain to the outside, because they are attracted by reelin which is produced by special cells near the brain's surface. The neurons' growth cones (1) sense reelin, (2) move along the fibers of glia cells and (3) end up in their final position.

Now you might ask how neurons finally connect. Well, from studies at the neuromuscular junction (synaptic connection between a motor neuron and a muscle) we know that initially the growth cones of many motor neurons are attracted to the muscle and connect to it. Then, when one of the motor neurons fires action potentials, it simultaneously releases a chemical. This chemical acts on the post-synaptic site and leads to the aggregation of the neurotransmitter receptors in the near of the axons that released that chemical. So the density of receptors near the axon of the neuron that fired is greater than the densitiy near the axons of the other neurons. So, when now another motor neuron fires, there are not enough receptors close by and, finally, all axons are eliminated but one. This is but one (very simplified) example for the formation of connections.

It might be interesting to know how you would model these principles in a neural network. As far as I know this has not yet been tried in this detail.

[phil123424]

Good point StriatumPDM, this kind of selection at the neuromuscular junction is an option to select neurons, which is also used in other terms f.e. the selection of the dominant follicle in the maturing of germ cells. Is that not an example for the variable opportunities of adopting, which is so immense important that it is might be a very basic function at the molecular cell level (conserved very deeply in the genetic code). So still my misunderstanding related to this field is based on the question, how could a artificial network offer such oportunities of adjustment. IS not a better understanding of phneomenons like this needed to go further in create artificial hypothesises? F.e. Why differs the reformation after damage of neuronal connections in PNS and CNS. Which are the factors which enables or disables the reformation, and why should the connections in the brain not be reconnected? But how does the brain manage to maintain or restore the function after loss of a great amound of cells, which is shown after hemispherectomy.

[drtom007]

One of the major differences between CNS and PNS axonal restructuring is glial scarring. In the PNS, there are large amounts of circulating macrophages that clear debris from the injury site, along with an intrinsic capability of Schwann cells to assist them. Schwann cells also secrete chemotaxic factors to assist in migration of new growth cones along the laminins in the extracellular matrix. In contrast, CNS lesions are littered with debris that is not removed due to the presence of oligodendrocytes and their precursors, microglia, and inhibitors of myelin formation which block access of phagocytic cells. Furthermore, the ECM in the CNS does not express the necessary laminin proteins to provide the proper scaffold for migration to occur, nor are growth factors re-expressed to stimulate axon growth. The result is a scar that cannot be cleared and obstructs migration.

Hemispherectomy is only used in children because their neural pathways have not been fully established yet. There is room for new connections to form and take over function from the lost tissue. This remodeling is not perfect, however, as all patients undergoing this surgery show at least some degree of partial hemiplegia. A hemispherectomy on an adult would have devastating consequences because they cannot remodel their established connections (plasticity) and would lose all function from the removed tissue

[phil123424]

Thanks DrTom for your reply, i think I mainly understood your detailed explaination of the scarring mechanism. What I really would like to know is why are there differences, when we know them could we use them maybe to treat at least spinal cord injuries, and maybe later on higher functioning areas. For example, give the environment of the brain the tools to create new streets for the growth cones, lead the mikroglia to clear the debris or attract macrophages into the cns area, use the factors or even the swann cells of the pns for cns repair by transplantate them. Which are the main borders to do that at the moment to use the abilities of plasticity also in the cns.
Because in my point of view there are abilities of remodelling in adult brain areas left, what you can see after insults with the proper therapy. The functions can be carried over by other pathways, when they have similar "algorythmic abilities". But what are here the critical factors?
Time, neurotransmitter existence, the function itself? F.e. I could imagine that the motory skills are very hard to gain back because the are early developed and kind of hard wired, without many connections to other functionary areas, in contrast to brocaphasias. Maybe its also important to know how to relax existing neural circuits in that kind of manner that the could take over the loosen function by creating a broader wired connectivtiy.