Theoretical Neuroscience.

Let’s discuss this in slightly more detail. Suppose that one single neuron can remember one particular signal and then be able to recall how to respond to it appropriately. That is to say, when there is a incoming signal, (stimuli), it is assigned to a free/open neuron that remembers the stimuli and, from then, only responds to that one stored stimuli.

In our theory, we propose the hypothalamus as a form of “factory” for the signal system. Many studies show that the hypothalamus is physically connected with virtually every single part of the brain. In regards to our theory, we will imagine that we have sets of neurons that are physically connected to the nerve endings of our signal system in order to place an order for the production of the necessary signals. As well as, sets of neurons that react to one of a hundred signals from the signal system, and neurons that react to external sensors.

In order to support our theory, we need to find a set of neurons that react to a wide range of hormones, but this is a task that is quite challenging with the current level of technology because the amount of these types of neurons is quite small relative to the total mass of an organism. We can assume that the number of neurons reacting to a particular hormone will be several thousand in the total mass of about one hundred billion.

Let’s try to imagine how this would work in context with our theory.

We take a single word, it doesn’t matter what word it is, but it is important to note that any word is an established pattern of sounds describing an object, event or state. When we put several words together, we form a new pattern which gives us a more detailed description. That is, within the framework of the theory, the relationship between the patterns will be the same at any level of abstraction. We can also say that these patterns will be grouped relatively close to each other.

Since it is somewhat difficult to consider and understand the proposed system through binary examples, it will be most clearly demonstrated through the example of structured information. For this, and piece of text is suitable. Let’s set a condition neuron equal to one word.

Below, you can see first visual model of this type of system.

As you can see, one set can consist of several other sets, which may be partially or wholly owned by other sets. Different colors represent different sets; it’s like a Russian stacking doll, except some parts of the doll can also be part of others at the same time and the level of investment can reach quite significant values.

Consider each group of neurons as a set, and the entire system as a relation of these set with each other. In order to describe such a system we must make use of diagrams like above. We can say that the graph is precisely the optimal structure that allows us to organize all the different sets of neurons.

The smallest sets will consist of neurons that are connected to the hypothalamus which recognize hormones and request hormone production, as well as sensory neurons that process external stimuli from their entry point and through to the brain and other sensors.

The set for motor neurons will look exactly the same, but instead of processing sensors, they will issue commands to the body to be executed.

That is to say, in the case of sensor sets of neurons, the incoming stimuli merges to the higher neurons and through the motor set, break apart into several neurons to the downstream neurons, which will directly affect the action.

We can assume that through a message into the signal network, we can turn the command to the implement the action either on or off. For example, when we first learn to read we begin by saying every word out loud but eventually we learn to read silently. If there is no implementation command in the message, there is an internal dialog between the sets. We must also keep in mind that patterns within each of the areas have a physical connection to other patterns.

Let’s look at some basic examples of how learning takes place from the perspective of our theory.

In the first stage, the signals from the sensors are divided into minimally known sets consisting of sensory neurons, which produce the corresponding markers. At the second stage, the markers reach the motor neurons and begin to irritate them, and if the stimuli do not belong to the same group, then it breaks the balance and causes the system to create a new set for the stimuli and assign a marker to it. The new marker will also be given a similar group of sensory neurons.

An excellent analogy for the process that we are proposing, is the way that we decipher forgotten languages. We look at the entire piece of text and look for repeating symbols and patterns and what context they’re found in each time. From this, we can begin to build an understanding of what has been written from the relation of the symbols to each other.

In the example above, we have a diagram of two words, one of which is written with an error. If our system continues learning this model, then eventually the word with the correct spelling will be used in others sets more often that the wrong spelling. It is also very likely that the correct set with take precedence over a neuron that combines a set with an incorrect spelling. If what we hear or see uses a significant part of an already familiar pattern, we can assume the general sense or correct spelling. With age, frequently used sets take on a higher priority due to the frequency of their use in larger sets. Presumably, this is what we call “life experience” and “character” in everyday life.

Similarly, the context of a conversation is remember for a short amount of time since the hormones of the signal system are not immediately removed from the blood, they continue to irritate the same patterns. This means that for words that have several meanings, the system continues to revert back to the previously activated set. In regards to the process that we call consciousness, we can assume that in many aspects this is a process that occurs between two sets of neurons using a signal network that strongly resembles an internal dialogue.

What is most important to notice about our system, is that the a set should always lead to some sort of action. This action, however, doesn’t necessarily always have to be a real action in the physical world. The completion of a set and realization of our interest and desire to achieve a certain goal, restores the balance to the system.

When working on our computer model of this system we encountered an interesting point in the process. Certain sets that were used to describe the system’s own internal content that were also found within other larger sets were having duplicate copies created of them, and the creation of these copies was constant. This is a process that we described as “deja vu”, and in our case, it lead to a problem with the indexing in the database that we were using.

It is worth noting that when we started this project, we did not have the task of recreating an exact copy of the biological processes taking place in the brain. We assumed possible evolutionary steps and on the basis of them tried to create a structure that would be capable of processing a lot of sensors and implementing the control of a large amount of muscles. During the development process, we created six versions of the program code, and implemented the possible structure of the sensory part of the neurons. Each version introduced additional data and aimed at clarifying the bigger picture. We can state that the model functions in a very similar fashion to the living brain.

It is worth pointing out that the signal system is a bit more complicated than the simple message transfer system that we’ve described. What we’ve written out can be imaginatively represented as a kind of broad stencil across a wide array of patterns.

The “sensor structure” created by the implemented part of the model can be used as a self-sufficient structure. One can imagine this as a forest, of sorts, in which trees share some branches and leaves or even pieces of trunks. That is to say, although these trees are in most cases separate structures, they intertwine and share certain pieces with each other. It’s a system in which it’s easy to see where certain pieces branch off in separate directions, yet where they also connect with others, and where their different places of origin exist.

At the given moment, based on experience with previous versions of our system which implemented only the sensor system, we’re in the process of putting together a full and complete version of everything we’ve discussed so far. The new version will implement the entire algorithm of the model, including a self-learning process. This system will have to identify a speaker and self-learn to synthesize words and phrases from the minimum tones available to the DAC.