Neural system: correcting errors

My last post failed miserably. And all because I made a serious mistake by building a multi-level abstract model. We will fix it.

So what was the problem? The answer is simple. I proposed a logical model, having built it using abstractions, but it was taken very literally. It turned out about the following:
- Let's imagine an electric current in the form of water, and an electric circuit in the form of a system of tubes.
- And the water must be distilled?
- It does not matter, we are only interested in its properties as a liquid.
- And what temperature should it be?
- This is also not important.
- What material are the tubes made of and how are they fastened together?
- And it does not matter at all, it only matters that they are able to conduct water.
- No, well, if everyone starts to build such models without having a scientific basis and authoritative evidence, then Habr will have to be closed. Minus you in karma!
Therefore, I repeat once again that you should not cling to the form. It is important to learn to see the content. Well, if in the process of reading it becomes completely incomprehensible, then you will need to return to the buckets and stones , and then try again.

And I really hope that this post will be read only by those to whom the topic is really interesting and comments will be on the merits, and not with the requirements to provide links to reputable sources.
')
There is no choice and will have to go a long way. Let's imagine that we are building an artificial intelligence system. What does it have to function in the same way as a person, that is, to exactly repeat all his brain activity, including consciousness.

To do this, redraw the image from the previous post and get the following.



Let's figure out what's what. Let's go from the bottom up.
At the lowest level, we have a basic set of programs responsible for system initialization, service functions (cleaning unused caches, rebuilding IDs, etc.), security and protection. There will also be continuous monitoring of all the main parameters of the system and in the case of unauthorized intervention, an attempt will be made to restore the integrity, and in case of impossibility - shutdown.

The next level is also part of the serving subsystem. It contains application programs that the system itself will create in the course of its work for its convenience. For example, she can create a program that will be responsible for the ability to synthesize speech in order to communicate with us.

Next is the decision making subsystem. She is responsible for the reaction to external influences. It happens as follows. Based on the input parameters, a quick search for a solution is made. To do this, use the cache. The local cache is the decisions that have already been made, the global cache is the ability to get a solution from the operator. If such a solution is not found or there is no possibility to use the global cache, then the input parameters are analyzed and a control signal is generated. That is, we have three functions f1 (p1, p2, ... pn), f2 (p1, p2, ..., pn) and f3 (p1, p2, ..., pn). The first one returns the answer from the local cache, the second one from the global one, the third one forms the solution on its own. All results are stored in the local cache. In this case, the more often a decision is made, the more weight it has. In this way, a “habit” is produced. It is obvious that the functions f1 and f2 will be the least energy consuming and the system will use them if possible, and only as a last resort the function f3.

Well, the most important part is the self-monitoring subsystem. In the background, it will check all decisions made and signal that some of the decisions were wrong, forcing it to reconsider and either reduce the weight or exclude it from the cache altogether.

This is a general view of the work. As we know, for the functioning of any system, three components are needed: the hardware level (hardware), the software level (software) and energy, without which any processes are unthinkable. We reviewed the program level. Now we need to deal with the hardware. Modern computer systems are built on the chips that provide the transfer, processing and storage of information. The processor is responsible for executing commands, hard disks store information, everything is transmitted over data buses in coded form with zeros and ones. Well, electricity is used as energy.

Now you can draw analogies. The brain is a hardware level in which there are areas that provide information storage, and areas that execute commands. It all works due to the same electricity, which translates the energy absorbed by us in the form of food. Well, our consciousness is the very program component.

Let us return to the question of where is the consciousness still in the brain or outside it? For this we use again analogies. I think that no one will argue with the fact that there is a physical level at which a program exists in the form of zeros and ones, and a virtual one at which it is already some kind of entity capable of functioning. The link is the program code. These levels are closely related to each other, but at the same time, at each of them the program has its embodiment. Exactly also with consciousness. It can not exist without the brain, but at the same time, brain activity is not just a set of neurons and connections between them, but a kind of virtual entity that can be considered separately.

Next, we analyze how the system will be able to learn and reprogram itself.

And finally. Do not rely on authoritative opinions when questions concern consciousness. After all, who, except you, can most accurately say the heat to you or cold, sad or happy, painful or pleasant. All the answers are inside, you just need to learn to see them. But in order to accept something new, you must first abandon the old views. In the days of Giordano Bruno, society was not ready to accept the new. The end of it is sad, but the example is indicative.