While brain research has uncovered many insights regarding the functioning of nerve cells, researchers are still groping in the dark on one essential topic: namely, how the neuron network organizes itself. At the same time, brain research is increasingly dealing with phenomena that reach into philosophy and thus shake up the conception of humankind.
"Neurobiology allows no other conclusions than that there is no superior authority in the brain, but that the brain is a distributively horizontally organized system that organizes itself, without a conductor," says the brain researcher.
20 years after the boom in neuroscience, disillusion soon set in in scientific circles. While some things have been decoded about the architecture of the human brain's anatomical circuits, such as which structures are responsible for which outputs and how neurons function, essential questions still remain unanswered. "There are quite enormous methodological and intellectual challenges that we are facing right now," as German neurophysiologist and brain researcher Wolf Singer recently explained in Vienna.
How neurons connect to each other
After all, it is not possible to conclude how neurons behave based on the knowledge of how they are, Singer said. Therefore, it is necessary to understand "how information is encoded in the interactions between neurons and how the complex dynamics of neuronal interactions then leads to the functions that we call higher cognitive functions, i.e., perceiving, deciding, planning, and finally the phenomenon of consciousness," Singer continued.
The central problem lies in the high complexity of the interactions. No organ in the human body is as complex as the brain. There are about 60,000 cells in one cubic millimeter of cerebral cortex. Each of these cells can be described as an oscillator; it communicates with ten to twenty thousand other cells, which in turn receive their input connection from just as many cells.
Combinatorial explosion of elements
Cognitive systems - artificial as well as natural - need very efficient mechanisms to recognize, encode and classify relations and relationships between components. The reason for this is that the diversity of the world is based on the recombination of relatively few components (more than 90 atoms make up everything that can be distinguished in our material world, 28 letters are enough to write the world's literature, etc.). So the essence is in the combination and relations between the components.
A classical solution how relations and relationships are coded and recognized was invented in the middle of the last century with the perceptron (after English perception, "perception"). In principle, this is a simplified artificial neural network, which in its basic version consists of a single artificial neuron with adjustable weights and a threshold. The model is based on having detectors at the input that respond specifically to a feature A or B and distributing the activities of these detectors to an intermediate layer, also called the hidden layer. These divergent and convergent connections on the neurons of this intermediate layer then in turn have connections to an element at the output and can now, by adjusting the transmission efficiency of these connections, ensure that C becomes active only when A and B are both the case. The perceptron follows a simple principle that leads to a combinatorial explosion of elements, because one would need such an element for any object that is distinguishable. "If you take into account the complexity of the world, it goes very quickly to infinity," Singer said.
This principle of the perceptron is realized in all nervous systems and is almost exclusively the principle on which artificial systems are based. These also accommodate a larger number of hidden-layers. Through adjustment, after many learning steps, one ensures that for a certain input pattern, an element at the output responds particularly strongly. The difference between artificial and natural systems is also the reciprocity of the connections, because in artificial systems the flow of information only goes in one direction, whereas the study of macaque brains shows that there are hundreds of thousands of nerve fibers between the brain areas that go in thousands of directions and form a densely networked system.
Self-organizing system instead of central authority
It is assumed that artificial systems were also built this way because of the belief that our brain is organized in a similar way and that there is an instance at the top of the processing pyramid that has all the information stored in its memory and supplemented by sensory signals at its disposal. It is up to this instance to interpret what is happening in the body and in the world outside, to draw conclusions from it, to make decisions and to structure future actions. But: "This intuition is radically misleading if one takes into account what neurobiology has brought to light," says Singer and continues: "Neurobiology allows no other conclusions than that there is no superior authority in the brain, but that the brain is a distributively horizontally organized system that organizes itself, without a conductor," says the brain researcher.
This is also illustrated by the arrangement of the different areas of the cerebral cortex: each small volume of the cerebral cortex has a similar structure, and different tasks are performed in different regions of the brain according to the same algorithm. They differ only in the fact that the visual cortex receives information from the eyes and other areas operate differently than, for example, in the auditory cortex, which "receives" information from the ears; the internal processes, however, presumably run quite similarly. Thus, a distributively organized system in which a variety of sensory and executive processes are constantly operating in parallel.
One can easily study the circuitry in humans as well. If you look closely at these circuits from a sample of cerebral cortex, you see that information arrives from the sensory system, hits cells, passes activity to cells in higher and lower layers, showing serial processing. But in parallel, there are other connections that pull in a horizontal direction, stimulate others in neighboring regions, and are stimulated in turn. These account for 80 to 90 percent of all connections in the cerebral cortex, while in the area where the world is coupled in, sensory signals account for only six percent of all contacts from fibers directly and indirectly connected to the sense organs. Everything else is an internal interaction. "So the brain is an extremely autistic system that deals with itself and couples the world in very loosely," the brain researcher sums up.
Perception as the result of an interpretative process
Perception as the result of an interpretative process
The representation of a thing or object in the real world can be described in the brain as a spatiotemporally structured cloud of activity. Complex natural systems can recognize separations and connections of objects; artificial systems have difficulty doing so. This perception is also disturbed in some neurological diseases such as schizophrenia. Patients cannot recognize which objects belong together and which don't. A solution for this would be to define the relations as temporal relations by making sure that nerve cells dealing with this figure synchronize their activity for a short timespan and thus make them distinguishable, so spatial and semantic relations are translated into temporal relations.
Perception is in principle the result of an interpretative process. This involves the storage of immense amounts of information and its recall within seconds. Every second, the eye changes its direction of gaze four times, which means that different images hit the retina. These can only be interpreted because they are compared with the respective prior knowledge and read out as a result. The reason why we can store so much information lies in the dynamics of feedback networks. A model of the world is stored in the distribution of synaptic efficiencies of these billions of connections. This is how the cerebral cortex seems to work.
For decades the study of the brain was linear and dominated by the structure of artificial systems in which information always flows in the same direction from an input layer to an output layer. If one would only keep following these paths, it was assumed, one would understand how the system works in the near future. The realization that things are "completely different" in the brain has caused much disappointment, he says. But for Singer, it also has "something euphoric," he recently told the APA. The technology for researching the systems is now available - for example, the possibility of using genetic engineering methods to observe the entire nervous system of zebrafish in real time.
Singer, however, would still like to understand in the course of his research life "what the cerebral cortex does to be so extremely efficient, and what the secret was why it has been scaled up so much in evolution," he told the APA earlier this year. It must have realized a principle that is ingeniously applicable to very different problem solving, he said, and "radically different from anything we know in artificial systems."
Wolf Singer studied medicine in Munich and Paris. He is director emeritus at the Max Planck Institute for Brain Research (MPI) in Frankfurt, founding director of the Frankfurt Institute for Advanced Studies (FIAS) and the Ernst Strüngmann Institute for Neuroscience in Cooperation with Max Planck Society (ESI), and scientific director of the Ernst Strüngmann Forum. His research focuses on neural bases of higher cognitive functions.