Vitaly Dunin-Barkovskiy, Professor, Honorary President and Founder of the Russian Association of Neuroinformatics.
My topic is approximately the same as the previous paper, but my argument is that an understanding of the mechanisms of the brain, or rather an understanding of the mechanisms of understanding, is evidently a task that mankind must perform quite quickly, in the near future, and here I present the project of “Global Future 2045” on the reverse engineering of the mechanisms of the brain. Please write down the website address, there we have created an Internet laboratory for reverse engineering of the mechanisms of the brain, named after David Marr.
How is the singularity under discussion and the reasoning of computers connected? One of the ways of determining singularity is the moment when the computer becomes as smart as a human being. In fact, this essentially happened several years ago, when a computer beat Kasparov, and on the other hand, the viewpoint exists that understanding may serve to overcome the truly negative consequences of singularity, which seem unavoidable. The creation of reasoning for computers is mainly proceeding along two paths: one is the pure synthesis of reasoning: we think about how to make a computer truly smart. And the second is the reverse engineering of the brain, in other words to see how it is built, and try to recreate it. I will discuss the state of the problem briefly, although the previous paper gave you some idea about this, and also certain features of the path that we have chosen.
Synthesis of reasoning is going full speed ahead, it surrounds us, and we all know that the Internet intrudes into the lives of every one of us, and sometimes we cannot even tell where our individuality is, and where the individuality of our computer community. Virtually the entire system of control is computerized and computer-dependent. Naturally, these technical systems do not yet have any precise understanding: they understand poorly, even when they ask us to speak in natural language, and there may be only individual words. Reverse engineering of the brain is well-developed at present, and there are dozens of projects underway in various countries. I will only list the most famous of them: for example, the “SyNAPSE” project of the US defense department that has already been mentioned here, the “Blue Brain” project and so on. I will show you one slide for each project.
There is the sensational “Blue Brain” project in Switzerland, by EPFL in Lausanne and IBM. This project creates beautiful, wonderful films with an enormous amount of resources and expenses, where you can see how complexly, beautifully and wonderfully the brain works. I would strongly recommend that you visit their web-site and take a look, you can download truly excellent excerpts from these films. There are skeptics, however, who believe that it will take a long time to progress beyond the demonstration of these films, but skeptics are always skeptics.
The next project is by the European “Human Brain Project”. It brings together dozens, if not hundreds, of research projects from all over Europe, and includes comprehensive studies from the physiology of individual cells to the creation of microchips and solutions to specific tasks in specific details. The project is too large to understand when, how and where it is going.
The “SyNAPSE” project is an abbreviation which evidently means that an attempt is being made to synthesize the entire brain, starting at the level of the synapses, and this is what DARPA is working on. There were reports that they were reducing financing, and moving to more specific projects, but the project is still posted on the website, and this means that this task has been set – to attempt to reproduce the brain in its entirety, using an according amount of resources to do so.
Next comes “Computational Neuroscience”, a significant part of what can be done here, and in Germany there is a network of neuron computations. It includes hundreds of researchers of professorial level, a huge number of students and graduates, and 47 large groups where various aspects of neuron computations are examined.
There are many projects in Japan, and the leaders include the Brain Science Institute, which is practically the Japanese Center of nuclear research – the RIKEN Institute. It’s a wonderful institute, where around 50 specialists work, half of them non-Japanese. I had the good fortune to meet the first two directors of this institute, Masao Ita and Sunochi Amari. The first is an outstanding neuro-physiologist, and the second is a neuro-theorist.
After RIKEN, I can mention the International (initially it was intergovernmental) Neuroinformatics Coordinating Facility. It is headed by Sten Grillner from Stockholm, who spent several months in the Soviet Union in 1971, from where he returned home with a large number of ideas, and also scientists, with whom he received the first Nobel Prize for neuroscience in 2008, the Kavli Prize. There are 14 countries working here, and also hundreds of organizations, attempting to discover how the brain works.
Now about our path. I must say that we have very few resources, and relatively few centers. We chose the path of brainstorming: we are trying to understand how the brain works, with the help of the only mechanism that is capable of understanding anything – the human brain. We have set ourselves the task of assembling a relatively large group of 21 people, in order to examine everything that is currently known about the brain, and systemize it all. We have named our group after David Marr, an outstanding theorist who died young, and who could serve as a model of what can be done to understand the brain. Two years ago, on the 49th anniversary of Gagarin’s space flight, we were able to gain an audience on a similar topic with the world’s №1 neuro-theorist, Professor Hopfield, where we discussed to what extent these approaches were possible, and what we thought about them. After we decided how this should be done, we announced it in Boston, and then in Washington (on the last day in Washington, there were two speakers invited from Russia – Konstantin Vladimirovich Anokhin, and then I delivered a paper about this project).
What is our approach? When I first began studying the brain, I was told: “Why are you trying to understand how the brain works by sticking electrodes into it? It’s quite obvious that if you measure the field next to a computer, you won’t understand anything about what goes on inside it.” But now, in the 21st century, we know that you can’t measure the field around your computer, buy you need to protect it, otherwise enemies will find out what you are thinking. Why is this? Because we know the principles of how a computer works, and so if you measure it, you will find out what goes on inside it. So the main method that we plan to use in our work is to understand the principles: in order to understand the brain, you must first of all understand the technical principles of how it works. This is the contents of the Russian project on reverse engineering of the brain.
We have certain hope of success, because it seems that we are starting to understand the formats of the data: how the material with which the brain works is represented within the nervous system. The example of Watson and Crick provides a good illustration of what the format of data for understanding the system is: when they understood that there were just four letters for transferring genetic information, and then understood that there were 64 intertwined words made from these letters, molecular biology began to develop very quickly. And we are starting to understood the format of data in neuron systems, the codes of the brain. This is a very rough, general outline of the path that we are planning to take. At present a group is being formed, we hope to create a group of 21 physicists, strange as that may sound. The previous speaker was also a physicist, because one needs to be able to use any instrument that is required for understanding the mechanisms that we encounter.
During the year 2013 we will analyze the experimental data about the mechanism of the work of the brain within the group, in 2014 we will outline and describe the principles of the work of the brain. Before the end of 2015, we hope to have a full diagram of the brain, and after that – yesterday Alexander Ivanovich Galushkin spoke, and told us that by the time that we have a diagram, they will already have prepared hardware, and there will be something to reproduce this diagram on. The assembly of a full-scale working model of the brain will probably take place no later than January 2021.
At present, as I said, the group is being formed, and one of the first events in the formation of the group was the summer school on computational neuro-studies in Sudak in June last year. I would like to ask those present to put me in touch with people they know who perhaps are not geniuses, but sufficiently qualified physicists, who could devote themselves to this task – please put them in touch with me, and me with them, and so on. That’s all, thank you for your attention.