Human brain theory

7  The consequences of signal divergence for neuronal    substructures

With the onset of signal divergence, a significant change in the structure of many substructures of the brain occurred.

It began insidiously. In the spinocerebellum, the nucleus olivaris expanded in area, reserve neurons became involved in signal transmission. The transmission reliability grew.

But the laws of signal attenuation on markless fibres - and all neuron nuclei have no myelin inside - resulted in extreme value coding. Suddenly, joint angles could be analysed more precisely. Movements could be controlled more finely.

In addition, the class of inverted signals emerged. These could well be regarded as a new modality. This had far-reaching consequences for the cord ladder system, especially on the first floor, the cortex.

Different modalities separate early in the course of evolution. This is how the different loops of the cord ladder system developed: the temporal loop, the parietal loop, the occipital loop and the frontal loop, which also received the output of the spinocerebellum. There were now two modalities here: the original signals from the motor trunk receptors and the inverted signals from the same receptors on the opposite side. Both represented different modalities: The original signals were on-signals, their firing rate increasing with the strength of the measured variables. The cerebellum signals, on the other hand, were off-signals, they emerged from the on-signals by signal inversion (monotonicity inversion). Thus there were two modalities, but they each belonged together in pairs in a point-to-point mapping. The ON signal belonged to the ipsilateral motor neurons, the Off signal to the contralateral side.

When the output of the spinocerebellum not only caused the inverse excitation in the nucleus ruber, but also made its way to the top floor (the later frontal cortex), the sensory side of the frontal cortex now received 2 signal modalities: ON signals and Off signals.

These neurons began to position themselves exactly on top of each other. Thus, input layer 4 of the sensory side consisted of two sub-layers: The lower, older, were the previous ON signals from the receptors on that side of the body. The upper, younger layer were the off signals of the opposite side of the trunk, which reached the olive of the opposite side on a diversion from nucleus ruber, moved to the spinocerebellum, were inverted by it and now formed the upper sublayer in the sensory part of the frontal cortex in layer 4.

Incidentally, this was the case with many modalities. All signals that also passed through the spinocerebellum and were inverted there could lead to the splitting of the fourth layer into an on-layer and an off-layer in the cortex. This was also true for all later gyri.

And in other loops of the cord ladder system or gyri of the cortex, this double layer formation also occurred. Here, there were partly other causes: On-off systems already developed at the receptor level. On-ganglion cells and off-ganglion cells developed in the visual system in the retina. The cause was the formation of band synapses in the receptors. These were permanently excited and were relatively inhibited by their own receptor signals, which resulted in the formation of an additional off-signal. Other receptor systems (olfactory, vestibular, auditory receptors, olfactory and gustatory receptors, ...) also produced receptors with on-characteristics and parallel off-characteristics.

Thus, among other things, visual double-layer formation of the on-off type also occurred in the visual cortex - and much earlier in the occipital loop of the cord ladder system - in layer 4 of the sensory part of the cortex.

However, because a division of a layer in the layer system results in the division of the corresponding output layer, layer 3 also divided into two sub-layers. Here, however, the coupling was so strong that each sub-layer of type 4 remained directly connected to its sub-layer 3.

Thus, on the sensory side, there was always an outpour layer 3-On above the inpour layer 4-On, but an inpour layer 4-Off above which there was an outpour layer 3-Off.

Thus, the output layer 3-On lay exactly between the two input layers 4-On and 4-Off. It was to be given a special role: It was the nucleus for the formation of divergence modules with vertical signal superposition.

Because when the neurons in this S3-On-Output layer also divided, i.e. increased their number, in order to include reserve neurons in this way and to increase fail-safety, the output of the two neighbouring layers 4-On and 4-Off was distributed to precisely these neurons. The reason was the signal relatedness of the On and Off signals, which already led to the fact that both input variants always formed the double layer in a point-to-point mapping. This created the divergence module on the sensory side. Such divergence modules can be recognised by the large layer thickness of the sublayers involved. Such a divergence module with vertical signal mixing was described in chapter 1.1.2 (brightness module) and in chapter 1.1.1 (colour module).

The same thing happened on the motor side, but here the neurons of class 5 formed the output neurons. Because the sensory signals had split into the two complementary On-type and Off-type signal classes, the output class 5 also had to split into the subclasses 5-On and 5-Off. Thus, the fifth layer on the motor side of the cord ladder system or cortex split into a double layer S5-On and S5-Off. Here, too, the older sub-layer S5-On was at the bottom and the younger layer S5-Off was above it.

And when the formation of reserve output neurons began on the sensory side in the output layer S3-On, three, four, 10 or 100 now arrived in layer S5 instead of two output axons. Each of them occupied an interneuron in layer 5, at the same level as the output neuron in layer 3. This also created a module on the motor side, but in which the many input signals converged on exactly two output signals. The input ended at the top at an output neuron of layer S5-Off and at the bottom at an output neuron of layer S5-On. Between them were dozens, hundreds, thousands or tens of thousands of interneurons of the associated convergence module of the motor cortex, depending on the stage of development.

But it is not enough to limit signal divergence and signal convergence to the cortex or the olivary nucleus.

All cortex signals leave the cortex on the motor side towards the thalamus, towards the nucleus subthalamicus or towards the substantia nigra pars compacta (for movement detection by the basal ganglia), among others. Many of these signals - such as those from the basal ganglia, arrive back in the thalamus to overlap with others. Signal divergence is passed along all these signalling pathways. Therefore, the neuronal layers in the substantia nigra, the striatum, the nucleus subthalamicus, the olive, the cerebellar nuclei and also in the thalamus are no longer single-layer cell layers, but consist of both the same input and output layers 1 to 6 as the cortex, but also of ON- and Off-type sublayers with intervening interneuronal layers that bring the diverged signals back together. This can be seen well in the example of the visual thalamus, the corpus geniculatum laterale.

The divergence and convergence modules thus continue in all neuronal structures. This is why the central nervous system of humans and many other species has such a complicated structure and is so difficult to understand.

According to the author, the signal divergence in the neuronal structures began as a vertical signal divergence. The output neurones were located between the input neurones of the on-signals, which formed an independent neurone layer, and the input neurones of the off-signals, which also formed an independent layer. These received input from both the upper on-layer and the lower off-layer. The signals propagated vertically. The on-signals propagated downwards via interneurons, while the off-signals propagated upwards. Therefore, this structure was a divergence module with vertical signal propagation.
It is precisely this type of module with vertical signal propagation that must be the oldest from an evolutionary point of view. We therefore find them in reptiles and birds, for example in the dorsal ventricular ridge (DVR) and the hyperpallium, and probably others in lower vertebrates.
It can be assumed that here the axons of the interneurons were at least initially still unbranched and that the classical cable equation for myelin-free axons applied, in which the signal attenuation increased exponentially with distance. In this case, the signals were minimum-coded because the transfer function was strictly convex. However, the signals were unusable for motor control, as this required maximum coding. Therefore, these motor signals generally had to be inverted by the spinocerebellum, which made them maximum-coded again and thus usable for motor control.
As the axons of the interneurons branched more strongly in the course of evolution and ultimately formed large, highly branched "axon trees", the attenuation no longer increased exponentially with distance, but much more strongly and with the square of the distance. Finally, the ion clouds in the nerve cell no longer simply had to spread out linearly within a tube (axon), but spread out over the surface or even in space, which caused the ion concentration to decrease significantly.
The quadratic attenuation caused the transition from minimum coding to maximum coding, because now the neuronal transfer function was suddenly strictly concave and had a clear maximum in the concavity region. This meant that signal inversion was no longer necessary in the previous spinocerebellum. However, the spinocerebellum no longer regressed and could not degenerate. It received far too much input and generated far too much output. It was often still needed, for example, to invert motor on-signals and generate off-signals from them so that the neuron nuclei could form their double layer of on/off signals. Many other modalities now generated both on-signals and off-signals independently, such as the sense of sight. However, the output of these nuclei was now maximally encoded and did not require any further signal inversion in the spinocerebellum, so that this part of it was actually superfluous.
However, a fortunate circumstance led to the development of the pontocerebellum from this part of the previous spinocerebellum. The reason for this was the onset of signalling divergence in the surface, which was added to the previous vertical signalling divergence. This led to the development of the large average neurons in the primary and secondary cortex areas, which descended via the nucleus ruber and the nucleus olivaris as climbing fibres into the previous spinocerebellum, so that the latter developed the function of the pontocerebellum.
This process, which was particularly advanced in primates, gave rise to the pontocerebellum's ability to learn. The primary cause was the development of the spatial shape of the axons of the interneurons, which were no longer simple, straight tubes but rather resembled the root system of plants, extremely finely branched and with a strictly concave transmission function. This resulted in maximum coding and the signal inversion of the previous spinocerebellum could be omitted. The cortical mean value system provided the climbing fibre signals for cerebellar learning.
In addition, the Purkinje cells in the previous spinocerebellum changed their function. They were no longer used for signal inversion, but received the climbing fibre signal during the transition to maximum coding, which served to record the joint angles. This marked the transition from the inverting spinocerebellum to the sensorimotor protocol module, which is described in chapter 14.7. This is one of the prerequisites for body awareness.
However, reptiles and birds were also able to utilise cerebellar learning. Their neuron nuclei with minimum coding sent the signals to the spinocerebellum, where they were inverted and were now maximum coded. They travelled via the cerebellar nuclei to the cortex. This projected via the bridge nuclei into the developing pontocerebellum and simultaneously delivered the mean value signals to the climbing fibre system. Thus, reptiles and birds also had a pontocerebellum.
The difference between reptiles and birds, on the one hand, and (higher) mammals, on the other, most likely lies in the ability of mammals to realise the maximum coding in the neuron nuclei in the first stage, whereas reptiles and birds first generated minimimcoded signals in the neuron nuclei and then had to invert these in the spinocerebellum. Only then could the pontocerebellum utilise its learning ability.
The inability to directly maximise coding was associated with the fact that no planar divergence modules could be formed. This meant that it was not possible to expand neuronal and, above all, cortical structures in the plane (area). However, this worked in the birds' favour. Their brains were more compact and therefore lighter, which also resulted in a lighter head skeleton. This weight saving enabled the birds to take to the air. Every gram less brought an advantage.
It should be investigated whether reptiles and birds can utilise the advantages of divergence modules with maximum coding at all, possibly in the visual area. In the motor area, vertical divergence modules with minimum coding appear to carry out the signal evaluation, and the signal convergence in these areas is also realised by vertical convergence modules (hyperpallium).
The advantage of maximum coding in neuronal nuclei is shown in Chapter 14.7 "Sensorimotor protocol modules", which deals with spatial vision.
In mammals, a separation of the on-off modalities occurred parallel to the signal divergence in the plane, at least in the higher species. As a result, the previous on-off double layers dissolved and both modalities were now arranged alternately in the plane (e.g. primary visual cortex). As a result, the four signals belonging to a joint with two degrees of freedom (see eye movements or arm movements) ended at the corners of a square lying in the plane. Due to the maximum coding taking place in the plane, the joint angles could now be analysed more easily. This on-off separation of the double layers, which were still present in the visual thalamus, also occurred in the visual signals and can be detected cortically. It is quite possible that this signal divergence in the plane or in space (divergence modules with spatial signal propagation and maximum coding) can also be observed in birds. The reason for this is the development of interneurons, which enabled lateral signal propagation in the area of the retina, which was then taken over by the visual cortex.
The use of maximum coding gave mammals a considerable intellectual advantage.

It only becomes recognisable when one forgets the neurons (at first) and only looks at the signals. Then the modular structure of the brain becomes apparent and one understands the elementary basics of its signal processing. Those who first want to fathom how the brain thinks will fail. As a brain researcher, he should leave this question to the philosophers and study the signal processing in the brain.

Here it becomes clear that the Human Brain Project's plan to precisely record every synapse in the cortex will never lead to the goal. But if we understand the structure of cortical divergence modules and convergence modules, we will understand 60% of the functioning of the primary sensory and motor cortex areas. We can add another 20% understanding when we add the workings of the basal ganglia and their contribution to movement and speed analysis. The remaining 20% understanding of the primary sensory and motor cortex areas will come from the feedback of the brain's averaging nuclei. Although I do not yet understand all the connections here, the mean nuclei seem to be involved in judgements about what is good and what is bad for the living being. And they seem to act in such a way that "good" signals are promoted, strengthened, preferred, while "bad" signals are avoided and lead to motor and mental aversion. The key to this - let us reveal it here in advance - lies in the additional (unspecific) back-projection of the various mean nuclei via their transmitter systems into the cerebellum. There, the learning of positive signals is generally supported by corresponding transmitter release, while in the case of negatively valued signals, the storage of the associated inverse off-type signals is probably promoted by other neurotransmitters from the mean value systems.

Thus, the mean value systems are the judges of what is beneficial for the living being and what is not. This leads to the emergence of moral categories in the intellectual realm.

Within these algorithms, the pain receptors have the task of recognising disadvantages for the living being and forwarding them to the mean value systems. The dopaminergic system, on the other hand - originating from the regulatory circuits for olfaction, digestion and food utilisation - seems to evaluate the positive aspects. Whenever a positively or negatively evaluated stimulus occurs together with other signals, such as motor signals, the motor response is linked to the motor stimulus by imprinting. Positive reinforcement increases synaptic coupling with the original on-type motor signals, negative reinforcement acts on the inverse off-type signals. If the same stimulus occurs again while it was previously negative, the motor counterparts that were activated by the off signals contract. This is why you no longer burn your hand on the stove if the negative experience has already been stored.

Here the usefulness of splitting modalities into complementary on-types and off-types becomes clear. They are a prerequisite for the dual reaction possibilities of the living being.

Hot or cold, sweet or sour, salty or bitter, but also good or bad are the endpoints of a dual scale of evaluation of modalities, the basis of which is created by splitting many modalities into complementary submodalities.