Human brain theory

ISBN 978-3-00-068559-0

Monograph of Dr. rer. nat. Andreas Heinrich Malczan

7  The storage module in the pontocerebellum

The cerebellum arose very early in evolutionary history and represents the beginning of the transition from bilaterians to chordates. The spinocerebellum even stands for the transition from aquatic to terrestrial animals, as it enabled a stability of the body by introducing co-activation of the motor counterparts, which also took gravity into account.

The development of the cerebellum was described in detail in my monograph "Brain Theory of Vertebrates". Of course, my description there is to be understood as a hypothesis, for which there is, however, a great deal of circumstantial evidence.

In order to understand the origin of the pontocerebellum, we have to go back to the beginnings of the development of the central nervous system of the simple, segmented bilateria. There, there was one sensory and one motor centre in each segment per side of the body. Class 6 mean neurons collected the available excitation with their larger dendrite trees and thus gained mean signals for the different mean centres.

With the development of neighbour inhibition, the mean value systems of the different segments also competed with each other, and many fell victim to the neuronal competition. What remained, for example, on the seventh floor of the segmented nervous system was the dopaminergic midline system of the substantia nigra pars compacta with the subdivision of the VTA as well as the formatio reticularis, which probably extends over several segments.

In the second segment, just below the thalamus, a mean nucleus was able to assert itself, which developed into the nucleus subthalamicus. It received the output of the class 6 mean neurons from the sensory and motor nuclei of the first floor, which became the cortex.

Its output served to control the various life support systems. For this purpose, its signals had to descend and thus reached (among others) the output nucleus of the primordial brain on the seventh floor. This nucleus was the nucleus ruber.

However, the nucleus ruber projected not only descendingly to the motor neurons of the trunk, but also quite early to the lateral change nucleus of this floor (this segment). This nucleus was the nucleus olivaris. The purpose of the projection was initially contralateral inhibition, because initially both halves of the body were in neuronal competition with each other. This was caused by inhibitory interneurons of the contralateral nucleus ruber, which were excited by the opposite side.

Later, when these inhibitory interneurons had separated into a new nucleus, the nucleus Purkinje, this caused the contralateral inhibition.

However, his signals eventually reached the reticularis formation. This was a mean nucleus in this floor. The inhibition of the mean signals by the Purkinje nucleus led to the signal inversion of these very signals. The neurons of the formate reticularis, which received the inhibitory input from the Purkinje nucleus, inverted it and sent it to the nucleus ruber, quite soon formed their own nucleus. This is how the first cerebellar nucleus was formed. From then on, the Purkinje nucleus can be called the cerebellum, and the cerebellar nucleus is considered a subsystem of this cerebellum.

The cerebellum now supplied the opposite side of the body with inverted but excitatory signals. The contralateral inhibition of the motor counterparts was replaced by co-activation, i.e. inverse excitation. Thus both muscles of a simple joint were tensed, one more, the other less. This made it possible to set any joint angle, even under external load, for example due to gravity. Only then could the (future) tetrapods leave the water and walk on land.

The cerebellum received its complete set of signals from nucleus ruber - this was the simplest solution. Since part of the signals came from the mean nuclei above, the mean signals from the nucleus subthalamicus also came to the nucleus ruber and from there via the olive nucleus to the cerebellum.

Now, in the course of evolution, the tendency arose in signals and neurons to order themselves according to modalities. Here, the neurons of the different layers were also classified as different modalities from a certain stage of development and participated in the process of separating the modalities.

Thus it came about that the signals in the olive, the nucleus olivaris, arranged themselves according to the nature of their modalities. The segmental division of the body was maintained. This is illustrated in the following figure.

Primitive nucleus olivaris with the mean area of the projection to the future pontocerebellum

Figure 43: Primitive nucleus olivaris with the mean area of the projection to the future pontocerebellum.

In the primordial nucleus olivaris, the neurons of signal class 6, which are assigned to the mean signals, began to arrange themselves in two stripes. These mean value neurons were strongly outnumbered. Class 6 mean signals arose in each neuronal centre of each segment. They too were able to cross segmental boundaries in the course of evolution. Thus they could reach all other segments both ascending and descending. Ascending, they also reached the uppermost segment, which became the later cortex, moved to the motor side and descended again. Thus, the cortex also had all these mean signals.

Descending, these signals reached the nucleus ruber, which, however, also transferred them to the olive. There, these mean signals ended at switching neurons, which formed a horizontal and quite narrow neuron strip as shown in the figure above.

The olivary nucleus was therefore a body model of the signal-producing receptors, it was segmented like the body. It was also bilaterally symmetrical, with one half representing each half of the body. When the body of the animals was still worm-shaped, it also looked worm-shaped. And for each half of the body, it also received the mean signals in each segment, which now came together in a small spatial area because they were in the minority. This created a small strip of neurons in nucleus olivaris that received the mean value signals of the different segments.

These neurons projected into the cerebellum and also formed such a stripe there on the left and right (i.e. for each half of the body). At that time, the nucleus olivaris and the cerebellum were still very similar, both were as worm-shaped as the body of the animals at that time. The pontocerebellum emerged from the slender stripe of the median signals. This is my firm conviction. Many reasons speak for it.

In the course of the expansion phases of the cerebellum, a mossy fibre projection arose in addition to the climbing fibre projection from the olive. The input of the moss fibres came from the cortex via the bridge nuclei, where they changed sides. Thus, the moss fibre input came from the cortex side, which also fed the climbing fibres. It must be considered a great advance that the cerebellum could now also receive the descending signals from the cortex cortex.

Here, the topology of the cortex was transferred to the pontocerebellum.

In the cortex there was the parvocellular subsystem, which on the sensory side was formed by layers 3 and 4. On the motor side, the parvocellular system consisted of layers 2 and 5. Layer 1 of the cortex was a pure input layer.

The magnocellular system consisted of mean layer 6, which was represented both sensory and motor.

We analyse the cortical projections to the pontocerebellum. The parvocellular system projects via the bridging nuclei to the pontocerebellum, where the signals supply the granule cells via the mossy fibre system, whose axons cross the Purkinje cells at right angles and contact them directly or via interneurons.

Layer 6 of the sensory and motor cortex projects to the mean nucleus of the thalamic floor, i.e. to the nucleus subthalamicus. There, large mean neurons integrate the excitation, with several neighbouring mean neurons of the cortex each projecting into a common mean neuron of the nucleus subthalamicus. The cortex area that projects via the cortical mean neurons into a common mean neuron of the nucleus subthalamicus will be referred to in this monograph as the cortex cluster. It is associated with exactly one mean neuron in the nucleus subthalamicus. Its second-level mean signal reaches the cerebellum via a diversion from the nucleus ruber to the olive and from there as a climbing fibre to the Purkinje cells.

Each Purkinje cell is assigned a climbing fibre that represents the signal mean of a cortex cluster. The class 3 cortex neurons project via the bridge nuclei and the mossy fibres into a cerebellum area in which - among other things - precisely this Purkinje cell is located. We will call this area the cerebellum cluster.

Furthermore, we agree to call the signals of the class 3 cortical neurons elementary signals.

Then the cortical elementary signals project into the granule cells of a cerebellum cluster, while their second-level mean signal feeds the climbing fibres of the cerebellum cluster. The first-stage mean signal is generated in the sixth layer of the cortex by the magnocellular pyramidal cells there. The second level mean signal is generated by the subthalamic nucleus.

The cortical mean signals gained strength in the course of evolution because the receptive fields became larger and the number of cortex neurons (in many species) increased almost explosively. This is particularly true in primates, but especially so in humans. Nevertheless, the pontocerebellum began its work in the service of vertebrates much earlier. In many simple animals, the cortex is still quite small, but their cerebellum already forms a prominent structure of considerable size.

As the nucleus subthalamicus performed further averaging from the output of neuron class 6, the mean firing rate of its output neurons increased more and more during evolution. Thus, every Purkinje cell - if its cortex cluster had active neurons - received a strong mean excitation. And in many head segments, neurons were almost always active because there were constant environmental signals of an olfactory, visual, vestibular or tactile nature.

Now, a long-lasting, stronger signal leads to overloading of the neurons. Therefore, it was favourable when signal pauses were inserted. These signal pauses resulted from an inhibitory projection from the striatum.

We assume that the mean neurons of the nucleus subthalamicus also projected into the dopaminergic mean nucleus of the seventh floor, because every signal should be considered in mean nuclei. Therefore, the substantia nigra pars compacta received these second-level mean signals. It treated these signals in the same way as all signals. They were switched to dopamine and sent to the striatum, where they terminated at and excited inhibitory striosome neurons. These striosome neurons were now excited as long as the substantia nigra excited them.

Now, in the course of evolution, an inhibitory back-projection also developed from the striatum to the substantia nigra pars compacta. Of course, the inhibitory signals needed some time to reach the substantia nigra from the striatum. During this time, the striosome signals were able to inhibit the mean neurons of the nucleus ruber, so that the associated climbing fibre signals came to a halt because they were inhibited.

However, when the inhibitory striosome signals finally reached the substantia nigra pars compacta, they were able to inhibit it completely. From this point on, the substantia nigra could not send action potentials to the striosomes. Once those that were already on their way had been used up, the striosomes no longer received excitatory signals and their neurons fell silent. This removed the inhibition of the nucleus ruber, so that the mean signals from the nucleus subthalamicus were allowed to pass from the nucleus ruber to the nucleus olivaris and thus to the Purkinje cells.

However, since at the same time the substantia nigra was excited by the nucleus subthalamicus, it again produced excitatory action potentials, which after a certain running time arrived in the striosomes and there sent inhibitory action potentials in the direction of the substantia nigra and in the direction of the nucleus ruber. Thus, the signal permeability of the nucleus ruber was constantly interrupted for a short time by means of the striosomes and then restored. In this way, the typical oscillation form of the climbing fibre signals in the pontocerebellum was created: a tetanic oscillation that is continuously interrupted by short pauses. In this way, an overload of the neurons involved was avoided.

Periodic inhibition of mean climbing fibre signals by the striatum

Figure 44: Periodic inhibition of mean climbing fibre signals by the striatum

This gave the climbing fibre signals in the pontocerebellum the following appearance:

Climbing fibre signals in the pontocerebellum - basic appearance

Figure 45: Climbing fibre signals in the pontocerebellum - basic appearance

A strong mean signal - always present where there was strong cortical activity - could cause the process of long-term potentiation (LTP) and long-term depression (LTD) in a Purkinje cell. This process occurred when the Purkinje cell received two types of input simultaneously:

·       Cortical input via the parallel fibres.

·       Tetanic (higher frequency) input via the climbing fibres.

This is because a tetanic climbing fibre input caused an increase or decrease in synaptic coupling in exactly those synapses that were simultaneously excited by the cortical input within a very short time (in the range of seconds). A remodelling process in the synapses caused a permanent change in the coupling strength.

The change in the coupling strength remained permanently. This was due to remodelling work in the cell itself, the start of which was triggered by the tetanic excitation. In this monograph, we will refer to the permanent change in synaptic strength as imprinting.

If, after imprinting, exactly the signal that was present during the process of LTP or LTD occurred, the Purkinje cell reacted with a clear change in its firing rate. In the case of LTP, it now fired much more strongly, in the case of LTP, much more weakly.

The cerebellum thus became capable of learning.

But how could it be controlled which parallel fibres participated in the LTP and which in the LTD?

To do this, we need to consider the construction of the cortex and pontocerebellum from clusters.

For the analysis of the topology of the cortex and cerebellum clusters, we assume that the axons of the granule cells already branched in a T-shape and aligned parallel to each other so that they crossed the plane of the dendrite trees of the Purkinje cells at right angles.

For the analysis of the signal projection from the cortex to the pontocerebellum we use the cluster model . Let a cluster (in both cortex and cerebellum) be a simplified square section of the cortex. A cortex cluster contains a larger number of class 6 mean neurons, which, in the sixth layer of the cortex with its large receptive fields, absorb class 5 or 3 signals and generate mean activity from them. In sensory cortex fields, it is the class 3 neurons that feed the class 6 mean neurons. In motor cortex fields - and each gyrus has a sensory and a motor part - it is the neurons of class 5 that activate the mean neurons of class 6.

A larger number of the mean neurons from an equally larger cortex area now send the axons of the class 6 mean neurons to a common mean neuron of the nucleus subthalamicus. This particular neuron is our grade 2 mean neuron.

Now we can define exactly what a cortex cluster is:

A cortex cluster is a cortex area whose class 6 mean neurons converge on exactly one neuron of the nucleus subthalamicus.

For example, in the primary visual thalamus, a cortex cluster could consist of 25 brightness modules with spatial signal propagation. Then a square of 5 neighbouring modules in width and 5 neighbouring modules in length would be a cortex cluster.

We assume here (hypothetically) that the output signals of the different visual modules are also treated like different modalities and that their mean neurons also split according to modalities (light/dark, colour, line elements, ...) in the nucleus subthalamicus.

These different modalities have their own interconnected areas in both the subthalamic nucleus and the pontocerebellum.

We now consider such cortical clusters and their properties.

Neighbouring clusters are about the same size among themselves. We can group neighbouring clusters into cluster groups.

A cluster group, for example, consists of 9 clusters and is also square. The cluster with the number 5 is the inner cluster, the others we call outer clusters. Clusters 1, 3, 7 and 9 are the corner clusters. We call the direction in which clusters 4, 5 and 6 are arranged horizontal, they are arranged in a row.

We call the direction of clusters 2,5 and 8 vertical, these clusters form a column.

The following figure shows a cluster group, the 9 clusters are numbered as described.

In the cluster group of the cerebellum, the parallel fibres may run from top to bottom when the lower figure shows the cortex and cerebellum cortex from above.

 

Cluster group in the cortex

Figure 46: Cluster group in the cortex

 

We call cluster 5 the inner cluster and the others the outer cluster oredge cluster of the cluster group. We can assign a signal to each cluster, so the signal S5 belongs to cluster 5.

Within a cluster, the signal neurons are distributed approximately equally spaced. These signalling neurons belong to neuron class 3 on the sensory cortex side and to class 5 on the motor side. We refer to their signals as cortical elementary signals. They project via the bridge nuclei to the pontocerebellum.

As an example, a cluster with 25 output neurons is shown below, arranged in 5 rows of 5 neurons each. The real number of signal neurons in the cortex is certainly many times higher in higher vertebrates. Here we are only concerned with the organisational principle of the cluster topology. Structure of a cluster of neurons in the cortex

Figure 47: Structure of a cluster of neurons in the cortex

It becomes clear that a cluster signal, for example S5, is composed of many elementary signals that can be arranged in columns and rows and indexed. If the output O5,2,4 belongs to cluster 5 in the second row and fourth column, and if the number 5 represents the number of rows and columns, the signal S5 can be written as a matrix (greatly simplified):

 

Matrix

 

Here, s5,k,l represents the elementary signal in the k-th row and the l-th column, which is delivered in cluster number 5 by the signal neuron in the k-th row and the l-th column.

 

We postulate a signal projection of the signal neurons of the cortex clusters via the mossy fibres into the granule cells of the cerebellum clusters and assume that this signal mapping is topologically faithful in the sense that the cluster number is preserved.

A special feature of the pontocerebellum must be prefixed at the beginning. The output of the pontocerebellum is normally zero. This is because the excitation of a neuron of the dentate nucleus by the signal arriving via the climbing fibre is cancelled out by the fact that the Purkinje cell is also excited by exactly the same signal. Both have an equally strong excitation. But the Purkinje cell uses an inhibitory transmitter, so it inhibits the output neuron in the dentate nucleus. Therefore, there is normally no output in the pontocerebellum.

A structure without benefits would certainly have been eliminated in the long run. So there must be a benefit. This lies in the fact that the parallel fibres can excite the Purkinje cells directly, but can also inhibit them indirectly via interconnected star and basket cells.

This is where the important aspect of signal relatedness comes into play.

·       The cortex neurons of layers 3 or 5 within a cluster are signal related. This signal affinity arises via two stages.

·       These cortex neurons each supply (usually only) one mean neuron in its receptive field.

·       This mean neuron supplies exactly one neuron in the nucleus subthalamicus with all other mean neurons of the cluster.

·       Signal relatedness is passed on in neuron chains in both directions.

·       Therefore, the cortex neurons of a cortex cluster are signal-related to each other because there is a common signal receiver in the nucleus subthalamicus.

In neuron chains, signal relatedness is passed on in both directions.

If neurons are not signal-related, they are in neuronal competition with each other. They then inhibit each other, usually with the interposition of inhibitory interneurons.

Therefore, parallel fibres that receive input from the edge clusters can inhibit the Purkinje cell in the inner cluster because they are in neuronal competition with each other. Inhibition occurs through intermediate interneurons, which we call star cells and basket cells.

And parallel fibres receiving input from the inner cluster excite the Purkinje cells of the inner cluster.

This can upset the delicate balance between the excited dentate neuron and the associated inhibitory Purkinje cell, whereupon there is suddenly output from the pontocerebellum after all.

By default, however, the Purkinje cell is excited by the cortex signals from its own cluster, so it still inhibits the output neuron from the dentate nucleus. However, since the dentate neuron also receives the input from the inner cluster, both excitation components cancel each other out and the output is the zero signal.

Only if the signals from the outer cluster arrive simultaneously from the cortex via the parallel fibres do they inhibit the Purkinje cell because they excite the interconnected star and basket cells. Then what remains in the dentate neuron is the excitatory component that comes from the inner cluster. There is an excitatory output to the cortex.

This is interesting from a mathematical point of view: only when inside signals and outside signals act simultaneously, an output is produced.

Inner signal + outer signal = output signal.

The inner signal does not generate an output, the outer signal alone cannot do this either. Only both together can do this.

This is a signal linkage via an ampersand relationship. Mathematicians call this a conjunction. This is how complex signals are created. Complex signals are therefore always composed of an excitation component from the inner cluster and an excitation component from the outer clusters.

But the pontocerebellum could do even more: it could also learn these complex signals.

Learning means that the cerebellum, which does not yet know a complex signal, reacts to it only weakly or not at all. After the cerebellum has learned this complex signal, it fires significantly stronger as soon as this complex signal is presented as input.

We will call the process of learning a signal in the pontocerebellum imprinting. This imprinting occurs through LTP and LTD, which permanently and permanently changes the synaptic coupling between the neurons involved.

The basis for the imprinting are the following organisational principles:

·       The topology of the cortex clusters and the cerebellum clusters.

·       The neighbour inhibition of signals that are not (sufficiently) related to each other.

·       The tetanic mean excitation of the mean neuron. It occurs in a cortex cluster when a subset of neurons in the cluster is sufficiently active.

·       The process of imprinting (learning) through LTP and LTD using this tetanic mean signal.

·       The process of de-imprinting (forgetting) with the help of a low-frequency climbing fibre signal.

Thus, the tetanic mean signal is the store command, comparable to the store command in conventional computers, while the low-frequency mean signal corresponds to the erase command.

How does signal relatedness occur? It is transmitted from neuron to neuron in neurons. Mean signals are derivatives of cluster signals. Therefore, there is a signal relationship between them.

There seem to be - according to my hypothesis - two types of ground rules in neurons:

·       Arouse the one who is not yet aroused (if he can be reached).

·       Inhibit the competition.

For example, the mean neurons evolved. Any neuron that was active while the mean neuron was inactive could dock to it and now excite it as well. Provided its axon was long enough. Thus, each neuron is signal-related to its mean neuron.

If, on the other hand, a neuron has an inhibitory transmitter, it can only make synaptic contact with another neuron if both are active at the same time. Therefore, a neuron also inhibits any mean neuron in whose signal set its own signal is already present. This is why dopaminergic signals also inhibit the mean neurons to which they themselves have contributed. But now let's move on to the pontocerebellum.

How do the four organising principles play a role in imprinting in the pontocerebellum?

We hypothesise a more complex form of LTP and LTD in the pontocerebellum, which results in an output neuron of the dentate nucleus responding significantly more strongly to a specific signal constellation. We refer to this signal constellation as the imprinting signal . We refer to the process of LTD and LTP as imprinting.

The imprinting signal originates from a cluster group of the cortex and may consist of two partial signals:

-        The inner signal Si from the inner cluster of a cluster group.

-        The external signal SA from the external clusters of this cluster group.

The inner signal must not be the zero signal, a sufficient subset of the cortex neurons from the inner cluster must be active. It forms the basis for the signal average.

The associated mean signal terminates at the Purkinje cell of a cerebellum inner cluster and at the associated dentate neuron.

The internal signal Si terminates at excitatory parallel fibres of the Purkinje cell of the cerebellum inner cluster and at the associated dentate neuron.

The outer signal SA ends at inhibitory stellate and basket cells of the same cerebellum inner cluster. The dentate neurons of the inner cluster cannot be excited by the signals of the outer clusters because the distance between the neurons is far too great.

If the output neuron of the dentate nucleus is to respond more strongly to the complex signal S = Si + SA, the Purkinje cell must be less excited by these signals.

This would require the following:

-        The inner signal is applied to parallel fibres whose synaptic coupling is reduced by the tetanic climbing fibre signal (LTD), for example from the coupling value 1 to the coupling value ½. Thus, the Purkinje cell is less excited by the inner signal.

-        The external signal is located at the star and basket cells, which inhibit the Purkinje cell in the sense of a lateral inhibition. The synaptic coupling strength may have the value k. The effect of the tetanic excitation by the climbing fibre signal strengthens the synaptic coupling between the star or basket cells to the Purkinje cell (LTP), e.g. from k to 2k. Thus, the Purkinje cell is more strongly inhibited by the external signal. As a result, it will be even less excited.

-        As a result, the associated output neuron in nucleus dentatus is much less inhibited, so that its output is much stronger.

-        Since the tetanic climbing fibre signal also ends at the output neuron of the dentate nucleus, the synaptic coupling between the active mossy fibres of the complex signal and the dentate neurons also (probably) becomes stronger (LTP), e.g. from 1 to 2. Thus, the output signal becomes additionally stronger.

-        The synaptic coupling values altered by the tetanic climbing fibre signal remain permanently at the new level, so that an identical moss fibre signal to that at the time of imprinting now produces much stronger output signals in the nucleus dentatus.

Thus, several effects overlap and cause the excitatory output neuron in the dentate nucleus to respond much more strongly to the complex signal than before imprinting, because this output neuron is more excited and less inhibited.

If the imprinting signal occurs again later, the response is much stronger. We refer to the imprinting signal after imprinting as the Purkinje cell's own signal.

Reimprinting by the mean signal is not possible in this Purkinje cell. The moss fibre signals have a significantly shorter transit time across the bridge nuclei than the climbing fibre signal across the nucleus subthalamicus, the nucleus ruber and the nucleus olivaris. This means that the Purkinje cell is already inhibited by the moss fibre input from the outer clusters before the climbing fibre signal arrives. This makes it ineffective, because inhibited neurons represent a kind of short circuit for excitatory signals.

In nature, we observe a tendency to form reserves. In the cerebellum, this is realised by the interconnection of Purkinje cells into Purkinje groups. A Purkinje group usually consists of up to three Purkinje cells that learn the same imprinting signal. This provides additional security to the system.

A Purkinje group can be recognised by the fact that it ends with a Golgi cell. All Purkinje cells in a Purkinje group project to a common excitatory output neuron in the dentate nucleus. The function of the Golgi cells will be described later.

Let us summarise:

By imprinting using LTD and LTP, the imprinting signal of a cortical cluster group becomes the intrinsic signal of the Purkinje cells of a Purkinje group of the associated cerebellum cluster. After imprinting, these cells react to the reappearance of their own signal with significantly reduced inhibition of the excited output neuron in the dentate nucleus, so that its output in response to the own signal is significantly stronger. This allows the pontocerebellum to learn and later recognise exactly one imprinting signal per Purkinje group in each cluster. The recognition output reaches the pontocerebellum.

Each Purkinje group can therefore learn exactly one imprinting signal together with its output neuron in the dentate nucleus and store it as its own signal by permanently and permanently changing the synaptic coupling strengths for this complex signal. In this respect, the synapses of the involved interneurons of the cerebellum are the material places where the signal storage takes place.

Initially, there was exactly one Purkinje group per output neuron in the subthalamic nucleus. In order that more than one complex signal could be learned for the associated cortex cluster, two different possibilities developed.

The first possibility was that a climbing fibre contacted several Purkinje groups by increasing its length. Along this now significantly longer climbing fibre, several Purkinje groups could be contacted, as well as several associated neurons of the dentate nucleus. We will call this solution sequential divergence.

In the course of evolution, a climbing fibre contacted several Purkinje cells by sequential divergence, their number slowly increasing in the evolutionary series.

This created a new problem that had to be solved. If Purkinje groups are contacted by the same climbing fibre and receive the same cortical signals via the granule cells, then they will all learn the same imprinting signal at the same time. This would be a waste of resources. This is where the Golgi cells come in.

How can the origin of Golgi cells be explained? Whose descendants are they?

There was also lateral inhibition between the class 6 signalling neurons in the cortex from time immemorial to enhance the contrast of the output. The class 6 mean neurons transferred this cluster competition to the class 5 neurons of the clusters. This neuronal competition between clusters was taken over by the mossy fibres and passed on to the granule cells. Therefore, granule cells that belonged to a cerebellum cluster inhibited the granule cells of the neighbouring clusters. The interneurons required for this can be interpreted as descendants of the inhibitory interneurons of the cortex, which realised the cluster inhibition.

Golgi cells are descendants of the inhibitory interneurons of the cortical floor, mossy fibre neurons are the descendants of the cortical signalling neurons . Granule cells are descendants of the mossy fibre neurons.

The lateral inhibition between the cortical signals in the mossy fibre system is carried out by inhibitory interneurons, whose derivatives in the granule cell system are the Golgi cells.

The granule cell input of a cluster excited the Golgi cells of this cluster, which in turn docked with their axons to the dendrites of the granule cells of the neighbouring clusters and inhibited them if they themselves were excited. This is how the mutual inhibition of the clusters was realised in the pontocerebellum. In the spinocerebellum and the vestibulocerebellum, there was an analogous development. But in the pontocerebellum, this algorithm made it possible to prevent the so-called multiple imprinting. This is explained in the following text.

The number of Golgi cells is somewhat smaller than that of Purkinje cells. We assume here, for example, that there are three Purkinje cells for every one Golgi cell.

Golgi cells are excited by the climbing fibre signal and inhibited by Purkinje cells. This is because the climbing fibre signal is the mean value of the cluster signals, and these excite the Golgi cell.

Put simply, Golgi cells and Purkinje cells are signal-related, so they make contact.

As an example, we consider a group of three Purkinje cells arranged one behind the other, which are contacted one after the other by the same climbing fibre. As a fourth neuron, the climbing fibre may contact a Golgi cell. The further course of the climbing fibre may not be of interest at first.

We assume that the three Purkinje cells are imprinted with an S1 imprinting signal. They may activate exactly one common output neuron in the dentate nucleus because of their identical output. Thus, there is only one common output neuron in the dentate nucleus for these three Purkinje cells. Thus, for each imprinted signal, there are two spare Purkinje cells; if one fails, the circuit remains functional. Only the failure of all three Purkinje cells leads to the loss of the learned intrinsic signal.

Purkinje cells of the same Purkinje groups each end with a Golgi cell and are activated by the same climbing fibre. They are imprinted identically. Their output converges on the same dentate neuron. This reports the recognition of the imprinted intrinsic signal to the cortex.

Each Golgi cell is synaptically connected to the climbing fibre, which contacts the Purkinje cells located in front of it. During imprinting, this climbing fibre also activates the Golgi cell with its strong climbing fibre signal, which is strongly excited and interrupts the signal flow to the granule cells. This means that the subsequent Purkinje cells receive less strong parallel fibre signals because an entire population of parallel fibres (the foreign signals) is effectively pinched off. However, without the influence of extraneous signals, the output of a Purkinje cell is zero.

After imprinting, Purkinje cells recognise the applied imprinting signal during the pause in oscillation of the climbing fibre signals, as it has become their intrinsic signal. The inhibitory pathway fall leads to a strong output signal from the associated neuron in the dentate nucleus.

This is where the inhibitory feedback effect of the dentate nucleus on the olivary nucleus comes into play. Each excitatory output neuron of the dentate nucleus also activates an inhibitory projection neuron with its excitation, the axon of which moves to the olivary nucleus and docks there precisely at the input neuron that generated the formative climbing fibre signal. This greatly weakens the climbing fibre signal so that subsequent Purkinje groups can no longer be imprinted with this complex signal. When investigating the inhibitory projection from the dentate nucleus to the olivary nucleus, we have already noticed the strange circumstance that all dentate neurons of a cerebellum cluster move to a common neuron of the olive and have an inhibitory effect there. Here we explain what task this solves.

The inhibitory back-projection of the dentate nucleus to the olivary nucleus serves to prevent multiple imprinting. Without Golgi cells and without this inhibitory feedback, all Purkinje groups of a cerebellum cluster would be imprinted with the same imprinting signal.

Now, the pontocerebellum was able to learn several, mutually different complex signals per cortex cluster by means of LTD and LTP, because it was possible to prevent the multiple imprinting of different Purkinje groups of a climbing fibre with an identical imprinting signal.

Each imprinted Purkinje group reacts with significantly less inhibition of the associated output neuron to its own signal. As a result, the output neuron of the dentate nucleus reacts much more strongly to this signal. It reports the recognition to the cortex and simultaneously excites a connected inhibitory output neuron, which interrupts the climbing fibre signal of the cerebellum cluster in the nucleus olivaris. Thus, all other Purkinje cells connected to the same climbing fibre lack the tetanic excitation by the climbing fibre signal that is absolutely necessary for LTD and LTP. No other Purkinje group on this climbing fibre can now be imprinted with the same complex signal. Multiple imprinting is actively prevented.

Now, if a climbing fibre contacted several consecutive Purkinje groups as well as the Golgi cells arranged between them during its propagation in the pontocerebellum, as many different complex signals could be learned as there were Purkinje groups.

In the pontocerebellum, as many different complex signals can be learned per climbing fibre in each cerebellum cluster as there are Purkinje groups connected to the climbing fibre derived from the associated this cortex cluster.

Reminder: The climbing fibre signal was obtained from the mean signal of the cortex cluster and the striosome signal derived from it and distributed to the Purkinje groups arranged in a row by sequential distribution.

The sequential divergence of climbing fibre signals may have been developed to different degrees in different species. The more different complex signals a vertebrate could learn, the more specific and adapted its reactions to complex stimuli became. This gave it an advantage over other species that were not as good at this.

A much greater increase in the number of Purkinje cells occurred through signal divergence in the nucleus olivaris. An input was distributed there (after a long evolutionary development) to very many output neurons of the nucleus olivaris.

In the course of evolution, a signal divergence occurred in the olivary nucleus in which the number of output neurons increased. Each input neuron transmitted its excitation to many neighbouring output neurons. Thus, the number of climbing fibres assigned to one and the same cortex cluster was gradually increased. In parallel, the number of Purkinje cells in the associated cerebellum cluster increased.

Each mean signal from the nucleus subthalamicus, which had previously contacted exactly one Purkinje cell in the associated cerebellum cluster, now split into dozens, hundreds, thousands or tens of thousands of climbing fibres by signal divergence in the nucleus olivaris and contacted an equally large number of Purkinje cells in the associated cluster. This number was further increased by sequential divergence, in which each climbing fibre contacted several Purkinje groups.

If instead of one climbing fibre there was a whole series of climbing fibres running parallel to each other, the same imprinting algorithm took place on each of them. If the first Purkinje group of the climbing fibre with the number 1 learned an imprinting signal, exactly the same imprinting signal was also present on the climbing fibre with the number 100. Therefore, the first Purkinje group on this climbing fibre number 100 learned exactly this signal. In general terms, one could say that all the first Purkinje groups on all the climbing fibres to the same input neuron in the olivary nucleus were imprinted in the same way with the first imprinting signal. Thus, there was multiple imprinting on a large scale. This would be an enormous waste of resources.

A neuronal competition had to be organised, which only one Purkinje group won.

The foundation was laid by the signal attenuation in the nucleus olivaris. Those output neurons that were located very close to the input neuron with the cortical mean signal were supplied with more input and had the stronger climbing fibre signal within the cerebellum cluster. Distance-dependent attenuation occurred. They were the first to be imprinted with a complex signal. Their imprinting occurred with a time lag. The other Purkinje groups had not yet been able to reach the imprinting threshold. But still several Purkinje groups were imprinted, even if the imprinting strength was different.

When this complex signal appeared again, these already imprinted Purkinje cells in the nucleus olivaris inhibited the signal-providing input neuron. This inhibited his previously tetanic excitation, so that his excitation became low-frequency.

Now, those Purkinje cells that were pre-imprinted with the first signal, but not yet end-imprinted, underwent a process of de-imprinting. As a result, the partially learned signal was forgotten again.

A low-frequency climbing fibre excitation reverses the change in synaptic coupling strength caused by LTP and LTD. This requires a low-frequency climbing fibre excitation (e.g. of 5 Hz) for a duration of about 5 minutes. This was shown in studies by renowned neurologists.

Thus, a Purkinje group in the cluster that reached the imprinting threshold first in time was able to reverse the changes caused by LTP and LTD in the others, so that they became unimprinted (free) Purkinje groups again. They could now learn the second, third and each subsequent new signal by simply repeating this imprinting process for the new signals.

The low-frequency climbing fibre signal took on the role of a deletion signal. This created a neuronal competition between the end-stamped Purkinje group and the only partially stamped Purkinje groups, which freed up the memory space occupied by the partially stamped Purkinje groups. In this way, the cerebellum was able to use its resources effectively and sparingly.

The output of the dentate nucleus - i.e. the pontocerebellum - moved headwards into the cortex into an association area and established a secondary cortex area of first-level complex signals there. This cortex area also had mean neurons that allowed clustering. The cortical projection across the bridge nuclei and olive established new second-level cortex clusters in the cerebellum, which now received complex signals as input and could learn new, higher-level complex signals. These moved back to the cortex and founded association areas of the next higher level. This process was recursive.

The only thing missing now was the possibility of transforming complex signals back into elementary signals so that the original input - which had "disappeared" in the cerebellum - could be restored, since it was needed for the previously assigned task areas. There were several ways to do this.

On the one hand, the axon of the dentate nucleus moved to a class 4 cortex neuron and was transferred there to a class 3 cortex neuron. Its axon moved to the motor cortex and there contacted all those cortex neurons whose elementary signals were assigned to the complex signal. In this way, the complex signal diverged back to its elementary signals. Whether this possibility is actually used is still uncertain at present. The likelihood that this divergence variant will actually be used is rather unlikely, because on the motor side, signal convergence was the predominant variant. One only has to think of Betz's giant pyramidal cells.

Another possibility arises from an inverse circuit to the pontocerebellum, which produces the elementary signals again from the complex signals. I searched for this inverse circuit for almost a decade and found it in 2012 in the cerebellum, more precisely in the pontocerebellum. I called this circuit the inverse cerebellum. Several times since then I have had to revise my previous findings.

Many will think that I am artificially constructing a complicated system here just to spread my ideas about how the brain works. However, there are three facts that are internationally accepted among neurologists. These three facts are sufficient to prove the existence of an inverse cerebellum. I simply list these three facts and explain their significance.

Fact 1:

The pontocerebellum projects into secondary cortex areas, which in turn project back into the cerebellum via the bridging nuclei, where the signals reach the granule cells.

Fact 2:

All cortical elementary signals project back into the nucleus ruber. The nucleus ruber in turn projects (also) via the nucleus olivaris into the climbing fibres of the cerebellum.

Fact 3:

There is only one wiring principle in the cerebellum.

 

If one recognises that the moss fibre projection and the climbing fibre projection into the pontocerebellum follow exactly one principle, i.e. that there are not two or even more differently working pontocerebellum areas, then one has already accepted the existence of the inverse pontocerebellum.

How can the work of the inverse pontocerebellum be explained?

The secondary cortex receives the complex signals from the pontocerebellum and sends them via the mossy fibre projection into the inverse cerebellum. These complex signals are the granule cell input in the inverse cerebellum.

The primary cortex sends its elementary signals to the nucleus ruber. This sends them to the olive. From the olive, these elementary signals reach the pontocerebellum as climbing fibre signals in the sections where the moss fibre signals of the complex signals form their own cerebellum cluster. We assign this cluster to the inverse pontocerebellum.

Thus, the elementary signals and the complex signals have reversed their roles. Whereas in the (non-inverted) pontocerebellum the mossy fibre signals represented the output of the cortical elementary signals, in the inverted pontocerebellum the climbing fibres are assigned to the cortical elementary signals.

Whenever an ensemble of elementary signals is active that have already been assigned to a complex signal in the pontocerebellum, as many climbing fibres are active in the inverse pontocerebellum as there are associated elementary signals. This is because every elementary signal that belongs to the complex signal is active and projects via the nucleus ruber and the olive into the inverse cerebellum. The activity of the climbing fibres sooner or later causes imprinting through LTP and LTD, in which the active complex signal excites precisely those Purkinje cells that represent the elementary signals of this complex signal.

As soon as the complex signal is activated, those Purkinje cells in the inverse cerebellum that represent its elementary signals are automatically activated. Their output reaches the thalamus and docks precisely at those neurons that are assigned to this complex signal. The thalamus projects these signals back to the cortex.

Once a complex signal has been recognised, it activates all the elementary signals assigned to it in the thalamus and the cortex. These activate the associated complex neuron in the pontocerebellum, i.e. the Purkinje cell of this complex signal. Its output in turn activates all the associated elementary signals in the inverse cerebellum. This results in a signal rotation that keeps this signal in consciousness.

Of course, this signal rotation does not last forever. As in the limbic system, it weakens as the firing rate decreases over time (saturation effects). And it is ultimately interrupted by lateral inhibition. As soon as other, sufficiently strong signals reach the thalamus in the area where the signal rotation is currently taking place, these will in turn activate inhibitory interneurons that weaken or even interrupt the signal oscillation.

During signal oscillation between the cortex, cerebellum, inverse cerebellum and the cortex, there is a constant change of the signal form. We distinguish between two signal forms: the elementary form and the complex form.

In the elementary form, the class 3 or 5 signalling neurons of the cortex are active and project to the pontocerebellum. The associated complex neuron - a Purkinje cell or the Purkinje cells of a Purkinje group - recognise this complex signal and respond with an output. This output is in the complex form. It reaches the secondary cortex area and projects from there into the mossy fibres of the inverse cerebellum, while the elementary signals project into the climbing fibre system of this area. The mossy fibres excite these Purkinje cells, whose output is again precisely the elementary signals. This completes the change from the complex form to the elementary form and the oscillation begins again.

This signal oscillation is observed in real terms in the human brain.

What benefits does it bring to people?

One benefit is the longer lifetime of a complex signal. It is stored temporarily in the rotation loops, so to speak. Some thoughts simply cannot be suppressed.

A much greater benefit is that automatic completion of incomplete signals now becomes possible in the brain.

We all know crossword puzzles. The aim is to find letters for which there are certain clues. Over time, more and more letters fill the rows and columns of the puzzle, but there are still many gaps in the letters, especially at the beginning. Humans are able to recognise the missing letters in words with gaps.

How does it work?

If a complete word is stored in a Purkinje group in the pontocerebellum, while the individual letters - each separately - on a moss fibre excite the Purkinje cells of an associated "speech recognition cluster" in the pontocerebellum, then for the word "bicycle", which is incomplete in the form of "F hr ad", each Purkinje cell will be activated, tapping into and being activated by the moss fibres of the partial letters f,h,r,a and d. The Purkinje cell associated with the word "bicycle" will recognise the word. The Purkinje cell associated with the word bicycle will recognise the word. However, its output is weaker than if the complete word were present. Each missing letter decreases the output of the responsible dentate neuron.

If there are many words with the partial letters f,h,r, a and d, there are several solutions. Then several Purkinje cells are active at the same time and report several solutions to the cortex. Depending on the match, with different strength. In our case, however, there are hardly any words other than the word bicycle. The solution is unambiguous.

The cerebellum system can therefore complete incomplete signals.

We recognise a horse even if we don't see it completely. The question "Who eats oats?" is enough to name the horse. But so do other animals that eat oats. And quite abstractly, we could even assume that a combine eats oats because they disappear inside it. Partial information is often enough to recognise the missing information.

If incomplete signals are understood as questions, the cerebellum system provides the answer. The answer springs from the subconscious: "A horse eats oats. "And now this subconscious has a name and a material representation in the brain. It is the pontocerebellum that provides the answer. More precisely, the inverse cerebellum, which brings about signal completion.

If one takes into account that the secondary cortex areas that receive the complex signals from the cerebellum are also only ordinary cortex areas, one can predict the further development. The complex signals activate cortex neurons, which in turn can be understood as elementary signals. For an elementary signal exists precisely when it is represented by the activity of a single cortex neuron. Thus, the first-level complex signals are practically second-level elementary signals. They project again via bridge nuclei and the mossy fibres into the granule cells of a new cluster in the pontocerebellum. This could also be called a second-level cluster.

Similarly, in this secondary cortex area, class 6 mean neurons form the mean signals, which in turn converge on neurons in the nucleus subthalamicus. Via the nucleus ruber, they reach the olive and the pontocerebellum. This is how the stage 2 cerebellum cluster receives its climbing fibre projection. The learning process - the imprinting of new complex signals of the second stage - can begin.

This process is recursive. Complex signals of level k become elementary signals of level k+1 in a new cortex area of level k. These cause the formation of new complex signals of level k+1 in the pontocerebellum. This continues until the number of involved granule cells, Golgi cells, stellate cells, basket cells and especially Purkinje cells is exhausted.

And in each of these levels, an associated inverse cerebellum arises in the pontocerebellum, which is able to transform the new complex signals from the complex form back into the elementary form. In each level, a signal rotation takes place in which the signal form constantly oscillates between the complex form and the elementary form of the respective signal.

Here, this process takes on a life of its own.

At the lower levels - for example, in the primary cortex - the elementary signals of the inverse cerebellum arrive in the thalamus and are usually suppressed by the currently arriving signals through inhibition. This is also due to the close proximity of these signals.

This is no longer the case with complex signals of the higher levels. Since they involve significantly more elementary neurons, which are consequently distributed over a larger area in the thalamus, they have greater distances between them. This reduces neighbour inhibition. If even different modalities are coupled together in the complex signal, lateral inhibition is impossible. At a certain level, the distance between the elementary neurons is so great that no lateral neighbour inhibition actually occurs. Then the signal rotation is no longer interrupted. Practically, the necessary interruption of signal rotation then takes place via the inhibition of the thalamus, which is controlled by the circadian rhythm of the nucleus suprachiasmaticus and takes place during sleep.

We are constantly aware of these rotating complex signals. And they are simultaneously present in the elemental form, so that we instinctively know what the horse looks like that eats the oats. And we know how it smells, how it runs, gallops, how it neighs. Not only the abstract concept of "horse" rotates in our head, but also - at deeper levels - all the characteristics we know. This is obviously a feature of our consciousness.

This is where the great weakness of the Human Brain Project becomes apparent. People who have never seen a horse, never heard of a horse, will not have the synaptic connections of the complex signal "horse" in the pontocerebellum and inverse pontocerebellum, thalamus and cortex. However, those who know horses have precisely these synaptic connections.

But while one person thinks a horse ploughs the farmer's field, another thinks horses are mammals, and a third thinks of horses only in terms of riding. So the synaptic connections of these people to the subject of horses are completely different. Especially the association areas of the brains of different people differ in the concrete type of synaptic connections. As much as their cultural circle, their level of education, their interests, their mother tongues and the knowledge they translate into complex signals differ from each other, the synaptic connections in the higher association areas of the cortex, in the clusters of the pontocerebellum, in the matrix of the basal ganglia, in the rotation loops of the limbic system and in many other substructures that still need to be recognised differed.

Only the primary cortex areas, which contain visual divergence modules for colour vision, directional vision, etc., have the same structure. But even here there are differences in the layer thickness, the number of input neurons and output neurons as well as in the concrete structure of the inverse modules. Therefore, the virtual replica of a brain with all its synapses is always only the replica of a special human brain, which is person-specific. Whether the huge amount of work is worth it and whether it can be transferred to other brains should be critically questioned.

The approach described here is much more promising. It can explain how thoughts, feelings, ideas and consciousness can arise in the brain.

Here, three signal stages occur in the brain, forming three subsystems. Each subsystem consists of an ascending projecting module and a descending projecting module that works inversely to the ascending module. At the lowest level, the signals are in analogue form. Each analogue signal, when active, is assigned a firing rate. We refer to this level here as the primary subsystem.

In the secondary system, the signals are present in extreme value encoded form. A neuron population of active neurons is characterised by a local excitation maximum that encodes the parameters.

In the third, tertiary system, all signals are in the complex form. This is explained in more detail in the last chapter.

For followers of neural networks, it should be briefly noted here that a cluster in the pontocerebellum is, in simplified terms, an associative matrix. The horizontal rows correspond to the moss fibres, the vertical rows to the Purkinje cells. The crossing points represent the synapses. The output of the cluster feeds a second associative matrix, but in which input and output are reversed: The output signals of the first associative matrix feed the horizontal rows of the second associative matrix, while the vertical columns of the second associative matrix take their input from the horizontal rows.

This represents a feedback loop in which each signal detected in the first matrix is used as input for the second matrix and is decomposed by it back into its original elementary signals. In this way, a signal rotation also takes place.

In this simplified model, the only thing missing is the integration of the climbing fibre signals in such a way that the usual and lengthy process of "imprinting" in neural networks is accelerated by mean value-controlled write pulses. A target function to control the imprinting process is then completely superfluous.


Monograph of Dr. rer. nat. Andreas Heinrich Malczan