2015-07-04

On neurons and human ideation

My kid is close to 4 years and started to exhibit a very human brain process: before sleepping, he starts to remember random ideas and expresses them. Sometimes they are not only memories and plans he had over day, some are really new and original ideas - even that many are yet childish of course.

It is not something that could be related to a sudden thought association, he just has "ideas". It does not happen only before the sleep, it just happens more often then. Before the sleep it is also easier to eliminate hypothesis like "this thing remembered him about that other thing".

This process could show some light about how brain and neurons works, and possibly helping to design more human-like artificial intelligence. I would speculate that the same process that makes ideas and memories to pop up out of nothing, to be the process that permits the most amazing thought processes in humans.

We know stories about people trying to solve very hard problems, and after a period of not thinking about it, the solution suddenly comes into their conscience. It looks very similar, no? Also, when we try to solve a difficult problem, random related ideas comes in mind without a clear pattern, leading usually to finding the non-obvious solution.

What can we tell about this process, in relation with what we know about the brain neuron's physiology?

Well, this external behavior leads to the hypothesis that there are neurons that stay semi-activated and that will fire later, triggering this sudden ideas into conscience. What is special is that this process seems to happen more often when the external stimulus is low. Somehow these semi-activated neurons are inhibited while the brain is engaged in processing strong external stimulus, then being slowly activated when there is no stronger activation in the brain (or maybe in that region).

The theory of neurons that are activated by other neurons that are firing does not model this behavior too well. We don't have that activity that would activate the semi-activated neurons more often when the brain activity is low, the activation should be even lower. We could have some automatic triggers that activates them (brain waves?), however this does not explain how the inhibition works.

My hypothesis is that there is a slow leakage of ions that slowly activate the semi-activated neurons without a signal from other neurons. Leakage is usually put in relation with a natural de-activation of neurons (losing potential), however we could imagine a process where the natural leakage would trigger the open of some pumps/gates of ion channels, similar to an activation from neuron's dendrites excited by other neurons. Just that this time the activation comes from "silence".

In order to implement such process very economical in the brain, it should be probably based on the change in the ions concentration outside the neuron's cell. Somehow a low activation around the neuron would create the chemical conditions to excite other semi-activated neurons. Maybe this is why the neurons don't use only one pair of ions (they are 4 ions if I remember well). One pair of ions might be used for neuron-to-neuron activation, while the other pair might control the inter-neuron inhibition.

Another process that could be involved is the synchronization of the neurons. We know the theory that "neurons that fire together, wire together". I would propose something a little more esoteric: neurons are activated based on time synchronization between different activation paths. This is based on the physical location of neurons and the speed of activation. Such process would make possible to have a kind of "matrix" activation, like when you select a row and a column in order to get the cell value from a memory matrix. However, such process in brain would work in much more many dimensions, because you could activate the dendrites of one neuron from many other brain locations that have the same propagation distance to the neuron. Also, the same activation can be created in various "meridians" of the brain, making possible to more complex links between concepts and also redundancy. This would mean that human memory would have a kind of holographic organization.

This process that I imagined is somehow similar to a theory about how the direction of the sound could be detected, even for simpler brains of animals: you propagate the signal through brain from one ear to another. If the sound comes exactly full-faced, the signal will overlap in the middle of the path. When the sound overlaps in another region, it means that the neuron delay difference correctly compensated for the sound delay between the ears. If the overlap is in the left, then the sound comes from the right, etc. The neuron(s) where the sound perfectly overlaps identifies the angle where the sound comes from, and the animal (or human) can turn the head to that sound.

This is only speculation for now. I am only profane in this domains. Please let me know if you have comments. Better, leave a message if you just exist, my dear reader :)

No comments:

Post a Comment