How to connect to the brain

Gary Styx, "In the World of Science" No. 2-2009

Basic provisionsFuturists and science fiction writers talk about the time when the brain will merge with the computer.

Cyberpunk science fiction, which appeared in the 1980s, habitually features "neuroimplants" that allow connecting computer devices directly to the brain: "I had hundreds of megabytes stored in my head," says the protagonist of William Gibson's story "Johnny Mnemonic." The main thing in this then-young genre (and at that time a megabyte seemed to be a huge amount of memory) was the comparison of outdated retroculture with technology that, as it seemed in the 1980s, did not go too far beyond the capabilities of a skilled biomedical engineer. Despite the fact that no one has yet managed to make such implants, either at the Massachusetts or California Institutes of Technology, the best cyberpunk authors created the illusion that these inventions would someday be made in reality and, perhaps, even during the life of the readers themselves.

Over the past ten years, quite convincing heralds of technologies invented in cyberpunk literature have appeared in reality. Today, a person with electrodes implanted in the brain can control an artificial arm with the help of nerve impulses alone, which gives hope to paralyzed people. Scientists are also exploring the possibility of sending electrical signals in the opposite direction, providing the monkey with the opportunity to feel how the robot hand controlled by it touches objects.

But how far can we go in creating spare parts for the brain and the rest of the nervous system? Will this technology ever be able to not only allow you to control a computer cursor or a prosthetic arm, but also turn 100 billion neurons of the brain into a repository of espionage information or into something else from the arsenal of Gibson's works?

Will a person turn into a machine?Today, Hollywood screenwriters and futurists, not so knowledgeable heirs of the cyberpunk tradition, are once again turning to such neurotechnologies.

The film "The Singularity Is Near", which is scheduled to be released in 2009, is based on the ideas of Ray Kurzweil, a specialist in the theory of computing machines, who stated that people will eventually achieve immortality by transferring the digital image of their brain to a computer or a robot.

However, the idea of eternal life in the form of a virtual character living on TV, as in the series "Max Headroom" (Max Headroom), or the transfer of personality into the newest humanoid robot today remains not much closer to reality than in the XVII century, when Rene Descartes reflected on the dualistic separation of soul and body. A full-fledged transfer of personality, as well as a computer simulation, for example, of delight from the play of sunrise colors, the changeable emotional palette of a person and the rest of the mixture of what together gives a completely subjective sense of the world and constitutes the essence of our conscious life, remain nothing more than an invention of science fiction writers.

However, behind the hype around prosthetics controlled by the power of thought, there is a lack of knowledge about the functioning of the nervous system, which is necessary to transmit information to the brain, allowing you to recreate cyberpunk virtual reality. "We know too little about the neural networks that provide higher cognitive functions," says Richard A. Andersen, a neuroscientist at the California Institute of Technology.

And if we approach it realistically, what can be achieved with the interaction of the brain and the machine? Do the first successes of experiments on controlling mechanical manipulators and computer cursor using EEG mean that we are inevitably and inevitably approaching, if not the Kurzweil singularity, then at least the ability to inject some high-level cognitive information directly into the brain? Will we be able to load "War and Peace" into our heads — or, as in the Matrix, a helicopter control manual? And how about writing down some words in a person's memory so that he doesn't notice it? Or should I pass at least one word there?

These questions are not only of academic interest. Even if the communication channel with the cerebral cortex will forever remain an invention of science fiction, then understanding how photons, sound waves, odor molecules, pressure on the skin turn into stable memories is no longer just entertainment for cyberpunk. Nerve prostheses built on the basis of such knowledge will help stroke patients and patients with Alzheimer's disease to find a new life.

Primitive devices for direct connection to the brain have already been installed in the heads of thousands of people. In severe hearing impairment, cochlear implants are implanted, which stimulate the auditory nerve with transformed sounds perceived using a microphone. Neuroscientist Michael S. Gazzaniga from the University of California at Santa Barbara described such a device as the first really functioning neuroprosthesis for humans. The electrode arrays that play the role of the retina are currently being tested in laboratories. If they work, then on their basis it will be possible to create devices that enable people to see in the dark.

More ambitious goals, like connecting the Internet directly to the hippocampus (the neural structure involved in memory formation), require technologies that have not yet been invented. To do this, we need to understand how stable connections are formed between neurons and the outside world, and find ways to translate the digital version of "War and Peace" into the language of interaction of neurons with each other. And hints should be sought in the works on the creation of brain-machine interfaces.

Your brain and textIn order to deliver the text directly to the brain, it is necessary first of all to decide whether it is necessary to insert electrodes into the brain tissue: after all, if this is necessary, then neuroimplants will have no practical value for anyone except the disabled.

However, it has been known for almost a century that the electrical activity of the brain can be detected without making holes in the skull. The device, similar to a bathing cap and studded with electrodes, can transmit signals from a paralyzed patient, allowing him to type letters or surf the Internet. Niels Birbaumer from the University of Tübingen in Germany, the leading developer of this technology, claims that magnetic stimulation of the brain through the bones of the skull in combination with an electrode cap to register the resulting activation will help to find out the localization of individual words in the brain. If you have such a map, you can aim to stimulate these points and form the appropriate memory, at least theoretically.

Some neurotechnologists believe that if specific words are placed at certain points in the brain (which is controversial in itself), then their detection will probably require better spatial resolution than what can be achieved with an encephalographic cap with electrodes. One of the currently ongoing experiments with implantable implants may be able to achieve the required accuracy. Philip R. Kennedy from the Neural Signals company and his colleagues have developed a device that registers the activity of many neurons. It allows stroke patients to send a signal interpreted by a computer as, say, a vowel, which can then be voiced using a speech synthesizer — and this is a step towards the formation of whole words. This type of brain-machine interface can also be used to stimulate individual neurons.

Even more precise devices should contain nanofibers with a diameter of 100 nm or less, which can be connected to individual nerve cells. Jun Li from the University of Kansas and his colleagues have created a device resembling a brush. The role of bristles is performed by nanofiber electrodes that stimulate neurons or register signals coming from them. Lee believes that this device will help people suffering from Parkinson's disease or depression, will allow you to control a prosthetic arm or even cause muscle contractions of astronauts during long space flights to prevent muscle atrophy in zero gravity.

How to learn a languageRealizing the fantasy of loading a textbook on higher mathematics (or even a phrasebook on French before a vacation) into your head will require a much deeper understanding of the brain signals encoding language and other higher cognitive phenomena.

The disclosure of the neural code is one of the main challenges to neuroscience, and to paraphrase Freud, this is the royal road to understanding consciousness. Theorists have put forward many different ideas to explain how billions of neurons transmit messages through trillions of synapses connecting them. According to the oldest of them, the code consists of the frequency of discharge of electrical impulses generated by neurons.

Despite the fact that the frequency code may be suitable for the perception of some stimuli, it is not enough to load the works of Marcel Proust or Richard Feynman into the brain, as well as to create a mental screen on which the "Madeleine" cookie from the novel "Towards Swann" or a conceptual abstraction from a textbook on differential calculus would be displayed.

In more modern works, attention is paid to the accurate measurement of time intervals between all spikes (time code) or continuously changing patterns of joint discharge of groups of neurons (population code).

Attempts to create an artificial hippocampus for people with memory loss will somewhat bring the possibility of such a download to the brain. This work should help researchers understand the coding process itself. The collaboration between the University of Southern California and Wake Forest University is aimed precisely at creating such a replacement for the nervous structure responsible for memory formation. The hippocampus, located deep inside the temporal lobe, is damaged by stroke and Alzheimer's disease. An electronic device that works instead of the hippocampus could restore a person's ability to remember new information. This project, funded by the US National Science Foundation and the Defense Advanced Research Planning Administration (DARPA), can go even further and provide an enhancement of normal human memory or help to understand the neural coding of high-level cognitive functions.

Two groups of scientists, led by Theodore W. Berger at the University of Southern California and Samuel Deadwiler at Wake Forest University, are preparing for publication an article demonstrating the ability of an artificial hippocampus to do the work of consolidating the memory of a rat, which was trained to press a lever to get a drop of water instead of a real one. Normally, the hippocampus sends signals that are transmitted to the cortical regions responsible for storing long-term memory. In this experiment, the rat's hippocampus was temporarily turned off using a chemical. When the animal pressed the right lever, an electrical signal from the sensory and other areas of the cortex was sent to an electronic circuit, which, according to the researchers, sent back to the cortex exactly the same signals that the hippocampus itself would generate in this case. Demonstrating that the artificial device mimicked the hippocampal output signal would mean an important step towards deciphering the neural code that could be used to write memory into the motor cortex; perhaps this would someday allow us to unravel the codes that underlie complex behavior.

If it were possible to figure out how a sentence or a whole technical document is encoded, then in theory it would be possible to transmit them to the hippocampus (or to the cortex) via electrodes, and the scene from the "Matrix" when the instructions for controlling a helicopter were uploaded to the brain using a cell phone would become a reality. "The US Department of Defense prefers a more prosaic example — the F-15 fighter control manual," says Berger.

The apparent simplicity of the signal transmission model in studies such as the creation of an artificial hippocampus mentioned above raises more questions than answers. Wouldn't such an implant erase a person's existing memory? Will the code for a certain sentence be the same for me, for you, and, say, for a person who speaks Kurdish? Will the entered code be able to easily integrate with other networks that provide the appropriate context, semantic structure of the sentence? Will there be an erroneous interpretation of the same words in some other meaning?

Some neuroscientists believe that the language of the brain will not be solved until we come up with something more perfect than registering neural spikes. "Experiments with the registration of multiple signals and subsequent attempts to figure out what these signals mean and how they correlate with specific behaviors will not solve this problem," says Henry Markram, director of the Department of Neuroscience and Technology at the Swiss Federal Institute of Technology in Lausanne. Each input signal received by a neuron or a group of them gives rise to a certain consequence, for example, the transformation of sensory signals into long-term memory in the hippocampus, acting in many different ways. "And since there are different ways, it means that we are still very far from understanding them," he says.

The Blue Brain Project, led by Markram, is an attempt launched in 2005 to simulate the work of the brain at the molecular and cellular levels using a supercomputer. To begin with, it is planned to create a model of a simpler rat brain, and only after that — a human. It will be possible to start solving the second problem only after the appearance of computing systems, the capacity of which will be 1 thousand. several times higher than the supercomputers that exist today. "I think there will be a conceptual breakthrough," says Markram.

The difficulty of not knowing the mechanisms of information transmission to the brain imposes a limit on how far neurotechnology can advance. The task of forming a set of connections that make up memory is strikingly different from magnetizing a group of bits on a hard disk. "Complex information of this type, like the contents of a book, will require the interaction of a large number of brain cells in large—scale areas of the nervous system," notes neuroscientist John P. Donoghue from Brown University. — Therefore, you will not be able to address each of them, forcing them to store the necessary information. Within the limits of modern knowledge, this is absolutely impossible."

So recording information directly into the brain can remain only a dream, lost somewhere in cyberspace. However, this does not deprive Donahue of optimism about other ways of transmitting information and creating controlled prostheses. He is a leader in research on implanting arrays of multiple electrodes into the brain, which allows direct control of a prosthetic arm and even a wheelchair.

Donahue predicts that in five years, brain-machine interfaces will allow a paralyzed patient to lift a cup and drink from it, and in the more distant future, such systems may improve so much that a person with a lesion of the upper part of the spinal cord will be able to do the unthinkable, for example, play basketball, embodying the events of the 1970s TV series G. "The Six Million Dollar Man" (The Six Million Dollar Man). Even without a channel for transmitting information to the brain, disabled people and researchers will reap the rich benefits of this simpler approach. Gert Pfurtscheller from the Graz University of Technology in Austria reported last year that a patient with a spinal cord injury could travel along a virtual street with just the power of thought. Miguel A. L. Nicolelis from Duke University, another pioneer in the field of brain-machine interface, began to study how monkeys connected to an artificial device controlled by the brain develop kinesthetic self-awareness, a sense of movement and touch, completely separated from the sensory input of their biological body. "There is some physiological evidence that during the experiment they feel more closely connected with the robot than with their own body," says the scientist.

The most important consequence of such studies may not be the creation of neuroimplants or mechanical hands. Understanding the processes taking place in the central nervous system may allow us to improve the methods of teaching children and better determine at what point to apply a particular pedagogical technique. If this is the case, then the research of neuroimplants and computer models of the brain will give more important practical results than dreams of using the brain as a flash disk, borrowed from the science fiction literature of the XX century.

Translated by B.V. Chernyshev

Additional literatureToward Replacement Parts for the Brain: Implantable Biomimetic Electronics as Neural Prostheses.

Portal "Eternal youth" www.vechnayamolodost.ru02.03.2009