展望神经

As usual, the main cause for progress in natural sciences are methodological developments. Of course, conceptual developments follow and go in parallel, but I think the, the tremendous expansion of the field of neurosciences was triggered by a number of methodological breakthroughs at the lowest level. These were the progresses in the domain of molecular biology, genetic engineering, which provided model systems for the study of development for the study of certain diseases.

These techniques also provided extremely valuable tools for the biochemical and molecular identification of neurons. We learned about a degree of diversification of morphologically seemingly identical neurons that are endowed with different molecular machineries. All this we owe to the development of molecular biology.

A second important factor I think was the development of non-invasive imaging technologies. cause this allowed us to transfer a large number of concepts that have been developed in animal experimentation to investigations in human subjects. So this is the functional magnetic resonance imaging pository emission tomography.

Also the introduction of the magneto ence holography, which allowed for much better spatial resolution for the analysis of electric activity in the brain led to a substantial expansion in the field. Then I think a non negligible factor was the availability, increasing availability of of massive computer power because as neurobiology started to apply multi-site recording techniques, which I consider essential for the analysis of neuron network properties, the data became increasingly complex and it became increasingly impossible to just evaluate the data by looking at them as in the classical times when I started in the field, there were no computers. We had to do cross correlation analysis by cutting long strips of paper and moving them over the floor.

So the availability of of computing power has greatly advanced at least systems physiology. I think it had a major impact on these developments. And then last but not least, the availability of these computer facilities allowed us also to treat properties of neuronal networks at a theoretical level.

The dynamics exhibited by neuronal networks are so complex that only modeling studies and theoretical approaches put us in a position to formulate educated hypothesis. Also, for the search of neuronal codes, we cannot proceed without these search tools. So I think altogether these methodological developments were responsible for the great expansion of the neurosciences, which I think is unprecedented in any Scientific discipline.

As far as my own Research is concerned, there are two parts in my biography. Initially I studied developmental processes in the brain. I was particularly interested in developmental processes that are shaped by experience postnatally.

And the rationale for this approach was that I felt that the organization of the mature brain is so complex that we better study the developmental principles and try to make sense out of them and try to extrapolate what these developmental processes will eventually lead to. But then an accidental finding that actually happened in this very laboratory where we are sitting changed my life because I discovered inadvertently that neurons in the primary visual cortex of kittens that were awake actually sitting on the table behind me watching a drifting grating, got entrained into highly synchronous oscillatory activity that I could not attribute neither to degrading to the stimulus nor to the main switch in Germany are 50 cycles because it was an oscillation in the 40 hertz range. And so for the first time I saw that spatially distributed neurons could coordinate their activity with very high precision in the millisecond range over fairly large distances.

And this triggered the idea that maybe time is used as a coding space for the coordination of distributed neuronal responses. And I found this option so interesting that I decided to neglect the so much beloved field of developmental neurobiology and invest most of our efforts into the further analysis of this phenomenon. And this is now nearly 15 years ago or more.

And since then we spent most of our time with multi-site recordings and the search for spatial temporal patents in distributed neural networks because it is our firm belief that information is not only encoded in the discharge rate of neurons as is classically assumed and confirmed, but that additional information is encoded in the precise temporary relations among the individual discharges of neurons whereby zero phase lake correlations. So precise coincidence is probably only the tip of the iceberg. But an important tip because synchronous activity is particularly interesting in neuronal networks because it propagates very easily synchronizing responses is a means to enhance the saliency of neuronal responses with great temporal position.

It can be done on a spike by spike basis. And the initial hypothesis was that this synchronization is used at peripheral stages of sensory processing in order to establish relations between features that are processed by spatially distributed neurons. Since then, many more labs have engaged in search for the synchronization phenomena.

It started I think in Europe with colleagues like RI who applied EEG techniques to assess synchronous oscillatory activity in the brain and relate these activities to cognitive functions. And since then the the role of oscillations and synchrony has been discussed in a very controversial way. Initial doubts that this phenomenon might exist, it all have now been countered I think, and we are now all engaged in the search for functional consequences.

And there is a whole plethora of hypothesis of what that could mean. Does the accumulation Of this vast amount of data that we are confronted with nowadays and that we have very great difficulties to cope with reflect true progress. It certainly does reflect progress because we get to know more and more facts and details about the organization of neurons, neuron networks and the brain as a whole.

And this is a necessary consequence of the classical analytical approaches that are applied in the natural sciences. One isolates a subsystem and then describes it in ever greater detail. But this is only one half of the scientific endeavor.

What we also need is the inductive strategy. This the synthesis of all those data. And what we are now confronted with is that we have a huge amount of words.

We have a tremendous vocabulary that is difficult to remember, but we are lacking the syntax to put those words together into meaningful sentences in many domains. This is the case in particular in systems neuroscience. Therefore we need, I think more theory because conditions have become so complex that near verbal descriptions are probably no longer sufficient to grasp or to accomplish this, this synthetic process.

We need formal descriptions, we need analysis tools that allow us to generate models and this requires very powerful computations. And then we need of course synthetic thinkers. That's probably the the most important ingredient in the moment.

People who can bundle a large amount of details into meaningful hypothesis or concepts. Now one can of course foster this process by providing the appropriate environment and worldwide one sees efforts to provide such environments by generating institutions that are usually called Institute for Computational Neuroscience or Institute for Theoretical Neuroscience. For the time being, many of these places are still existing in two great in isolation and are not sufficiently linked to experimental institutions.

Ideally one would have, one would like to see a very close cooperation between theoreticians and experimentalists, as has been so fruitful in physics where theoretical physicists and experimental physicists very closely collaborate freely, exchange data ideas and do this without jealousy, which is still not the case in our field. But I think biology is now approaching a threshold where it is mature enough to support theoretical approaches in a meaningful way. And here in Frankfurt, we have founded the Frankfurt Institute of Advanced Studies exactly for that purpose because I experienced that physicists and mathematicians, theoreticians in general when embedded in the experimental institute, they tend to to be marginalized.

They don't have enough colleagues to talk to because biologists usually are not trained in their language. And so the idea was let's create an institute in which several model systems, biological model systems that have in common to be multi-component systems that are tightly coupled like the moon system or the brain are studied by theoreticians, hoping that these theoreticians might discover common principles in these complex self-organizing systems and that they can share analytical tools, can share tools for model building. And I think it worked out because within about a year this institute grew from zero to 120 scientists.

And the special structure of this institute, which is not surprising to colleagues in America, but this is quite exceptional in Europe, is that this institute is funded exclusively by donors private funds. It is not a university institute, it is linked to the university in Frankfurt, but it is finance in a different way for the simple reason that it was impossible to install such an interdisciplinary enterprise within the conventional faculties because what happens there is truly I the disciplinary, it crosses faculties and they seem to be realizable only by having a erect and an own building and gather people who come from very different disciplines under the same roof. Even though those Terms have been taken from physical systems initially, they still grasp some of the characteristics of of of the brain because the brain clearly is a self-organizing system.

There's no conductor in the brain to coordinate all the distributed processes. The brain develops in a self organizing way through interaction between genes and environment. There is no instructor, there is no blueprint, a priori.

And finally the term applies very well to what we observe in complex systems all the time, namely that complex systems tend to exhibit properties that cannot be described or inferred from the properties of the components. And the brain is a particularly nice example for this fact, which we know from all complex systems there's nothing special about the brain. Something that is special about the brain is that we have on the one side material processes, neuron interactions that can be described from a third person perspective as any other process in our environment.

And that emergence from these neuronal interactions are mental phenomena, psychological phenomena such as emotions, thoughts, decisions. So there are realities which I would like to address as social realities which emerge or come into being by the interaction between brains that communicate with each other, mirror each other, ascribe properties to each other, could replace brains by persons, doesn't really matter because what makes the person is finally the brain in its in its bodily setting and due to these interactions phenomena emerge that are to be taken as real phenomena, they are realities, but they have to be described in a different language as the phenomena that bring them forth that are at their origin, namely neuronal interactions. And therefore I think that this is a classical case of emergence where something new is generated by a complex system, nobody would ever be able to infer from the knowledge, complete knowledge of neurons, and even from complete knowledge of neuron networks, that they are capable of developing value systems when they interact With each other.

I think the main Problem in our efforts to repair brains, apart from the fact that neurons do not easily regenerate in the central nervous system, they don't like to extend processes once they are severed for reasons that we do not know. Maybe there was no Evo evolutionary pressure on the development of such mechanisms because brain damage that really leads to cell death in in a large scale was usually not compatible with survival. The great problem in restoring neuron networks that have been damaged is that it is difficult to recapitulate the developmental process.

The mature brain is characterized by an extremely sophisticated circuitry that has developed step by step in a complex self-organizing process, guided by the dialogue between the genes and the increasingly more differentiated cellular environment that changes throughout development. And this process is a essentially bottom up process. Peripheral structures mature first, then they go through critical periods and the connectivity becomes fixed so that the, the higher stages can rely upon what has already been built and then they organize themselves as a function of what's already there.

And this process goes on in human beings until age 20 and then the circuitry crystallizes and then we have to live with the brains that we have got at that point of time. There's further modification by learning and neuroplasticity, but this is limited. Now if one were to replace neurons at higher levels of processing, for example, by injection of stem cells or so these neurons we know already, they can differentiate, they the precursors can become neurons, but the problem is will they find the partners with which they have to interact?

Will the information still be there That was available during development that guided the axon growth, that guided the formation of specific connections. And we know that to, to a large extent, this information is no longer available because the genes that one needs in order to express the necessary marker molecules, they get switched off once development comes to an end. And what is probably necessary is that one not only supplies cells that can differentiate into neurons and of course also ggl cells, but that one also switches on or helps the organism to switch on the genes that have been active during the developmental process that lead to the expression of all the molecular machinery that is required to tell the cells with whom they should connect, where they should connect.

And what of course cannot be reinstalled easily is all the idiosyncratic wiring and the adjustment of synaptic game that is due to learning processes. The newly integrated neurons will of course have no souvenir, no recollection of the experience that has been imprinted into the morphology and the structure and the function of neurons that have disappeared. So there are certain limits to the ability to restore.

And as we realize in the moment, there are certain places in the brain where this seems to pose less of a problem because nature has foreseen lifelong regeneration of neurons and apparently kept the machinery alive that is needed in order to promote this ongoing regeneration and insertion of neurons as in hippocampus and in the olfactory bulb. But the other structures where this is probably causing great problems, the first replacements that have been attempted and have had certain success even though it was not satisfactory yet, where the modulatory systems where topological specificity is not that much a problem. These systems are functioning more like endocrine or paracrine systems that liberate certain transmitters that are required in order to keep a certain tous of excitability.

These systems are more easily replaceable because specificity is less of a problem, but the future will show how far we get there. Certainly very promising approaches in reinstalling oxon outgrowth in several spinal cords by inhibiting factors that inhibit outgrowth in the adult and by activating mechanisms that come to action during development and promote outgrowth. There are encouraging successes in this field, but one should also consider very seriously the option of neuronal prosthetics to record neuronal activity from places where it is still available and then through mechanical devices and computer interfaces activate actors that can translate this activity into action for the motor system.

Or conversely, on the sensory side, cochlear implants already work admirably well. There are now attempts to develop retinal implants in order to reinstall rudimentary visual functions in people who have retinal degeneration. There's also a very promising option to insert light gated iron channels into retinal cells so that cells that are normally not light sensitive, which are the ganglion cells for example, or the bipolar cells become light sensitive so that one could in principle reins store some function that is lost when photoreceptors degenerate by simply transferring light sensitivity to second or third order neurons.

This is already possible in animal experimentation. It might be another promising avenue to pursue. I think Europe has a very strong tradition in the development of concepts end of 19th century helm those seminal figures who have provided the grounds for later experimental neurobiology to some extent.

They also are the grand grandfathers of a substantial number of systems physiologists who have been active in Europe. I think there is a good tradition in systems neuroscience in Europe has always been in the times when, when linear systems theory was still at its peak cybernetics, as it was called Europe, had a few very strong exponents such as rehad for example. Then the, there was something like, not iconic turn, but the molecular turn in Europe as well as anywhere else, molecular biology took over and replaced many of the systems oriented approaches.

Partly because initially it had been cheaper than to work with whole animals, partly because non-invasive technology was not available and one couldn't do certain things in animals, which can now be done in human subjects. But I think there's a strong renaissance in systems physiology even though in Europe we have to fight a hard battle against animal protectionists concerning primary research, this is a real critical issue. The labs that can, they still get the license to work with primates become less and less and those who still have the license.

This is the case with us where we have, we are exposed very much to public critique, but still systems neuroscience is having a renaissance. And I think there are good reasons for that because we will need it more and more in order to make sense out of the plethora of data that are collected by the properties of neurons and small circuits. And these data can only be integrated if we have concepts derived from systems neuroscience on the nature of the neuronal code, on the cooperation of areas in the brain, on global coordination of processes, on distributed processing.

All those questions are heatedly debated in the moment and we just need more experiments along those lines. And I think Europe is not in a bad position to do this. Israel as well, one should not forget the contribution, also conceptual contribution of our colleagues in Israel, which is substantial.