Tying Shoelaces with Artificial Hands

How can we establish a connection with a completely immobilized person? Why do we need brains and is it possible to create a prosthesis for the mind? These questions are answered by Alexander Alekseevich Frolov, Prof., DSc, PhD, Chief of the Mathematical Neurobiology of Learning Lab at the Institute of Higher Nervous Activity and Neurophysiology of RAS.

2045: Shortly before the New Year, Raymond Kurzweil, a famous futurologist and inventor, predicted that over the next twenty years people will be able to get completely artificial bodies. You work with “brain-computer” interfaces. Do you think there are solid reasons in your field of science to support this forecast?

Alexander Alekseevich Frolov: I think the question is invalid. Perhaps “to get” should be replaced by “to have”. “To get” can be understood like a statement that a human is something external to its own body and that a human can get a natural body as well as an artificial body. In fact a human is its body. And the question is whether a human can have an artificial body. And whether someone that has an artificial body can be called a human being.

I think the majority of people would agree that preservation of the personality is connected with the preservation of memory and, thus, with the preservation of identity. If the information will not be lost while replacing the natural memory device with an artificial one, the person apparently can be considered a human. And the same person as before the operation.

2045: Is it possible to conduct such an operation on a man in the near future, for example, in the next 20 years?

A.A.F.: I can offer no time estimates, but I see nothing that could fundamentally limit the possibility of such an operation. Already, surgeries are carried out that allow for an implant of artificial sensory organs in order to restore sight or hearing. Work is being done by Professor Berger’s lab at the University of Southern California to develop a prosthetic hippocampus. The hippocampus is unquestionably a center of memory, albeit short-term memory. The range of human organs that are open to prosthetics is constantly increasing, and even includes certain areas of the brain. The problem of creating artificial memory devices capable of storing the contents of the natural memory of a given individual is, understandably, complex but by no means unsolvable. There are a number of reasonably realistic approaches that could provide fruitful starting points for this line of research.

2045: Could you please explain what the term ‘“brain-computer” interface’ (BCI) means?

A.A.F.: BCI is a direct communication between a brain and a technical device. When any intention appears in the brain it generates a corresponding signal. A BCI is able to classify these signals and recognize the class of the signal according to the intention involved. Recognizing the class of the signal is equivalent to recognizing the intention. The intention is transferred directly to the device.

2045: Which achievements of the BCI would you call the most significant?

A.A.F.: The research of the BCI has already overcame the stage where it had to prove its importance on the laboratory scale. Now research into wider applications is needed.

2045: Has BCI found a practical application yet?

A.A.F.: An invasive form of BCI contains an electrode array which is implanted in the head, usually on the top of the cortex. They register a single neuron’s signals of activity at a resolution high enough to recognise the intentions of the brain. But this requires brain surgery. Therefore the invasive BCI can be used only on the completely paralyzed, for whom the BCI serves as the only means to convey their intents. Dozens of operations of this kind have already been done, and they showed their effectiveness. Meanwhile, further experiments aimed at improving BCI are underway, mostly using monkeys.

Non-invasive BCI uses the signals collected from the surface of the skull (the type of electrical activity recorded by an encephalogram is registered), or even at a distance of a skull (the type of electrical activity recorded by a magnetoencephalogram is registered). The source of the signals is the electrical activity in the brain. EEG are the most likely to see widespread applications; they are already used in the computer games industry as well as in clinical practice. Special ‘caps’, which do not require special settings, are used. These caps will soon become common household appliances like a TV or a PC. The skill of conveying one’s intentions to an external device by brain’s signals may itself be a thrilling game.

As for the non-invasive BCI’s clinical applications, the research is related to motor recovery in post-stroke patients.

2045: Returning to the question of an artificial human body, when do you think the BCI technology will become developed enough to perform the task of controlling such a complex thing?

A.A.F.: I have already mentioned motor recovery in post-stroke patients, which I think will open up wider applications of the technology. You know, creation of an artificial hand is by no means a simpler task then the task of controlling it via BCI. The point is that control of modern robots and robotic limbs is based on totally different principles than control of living bodies. That’s why those patients who moved their hands with the help of these manipulators considered them unfriendly and refused to use them. They felt unusual stiffness in the joints. The prosthetics are set to overly large coefficients of feedback, which eliminates deviations from the desired trajectory by placing additional resistance on the joints.

A human hand’s feedback network coefficients are smaller by orders of magnitude, which makes its movements slow and smooth. So, good mechanical models of organs and appendages are yet to be created. Next, the organs need to be sensitized, ie, given external and internal receptors similar to mechanoreceptors and proprioreceptors in a human hand, which will report about one’s hand and fingers position. The information these receptors provide may be converted into auditory signals, for example. The high learning capacity of the brain gives grounds for hope that after prolonged training these signals will be perceived as signals of the movement by the brain.

This process is called learning for sensorimotor coordination. Sensorimotor coordination is the basis of animal and human behavior. I expect a human to perform compound motions (to tie shoelaces, for example) with two hands: an artificial one and a real one. When one artificial hand is integrated, a second one may be added, etc. It will significantly expand the motor repertoire which may be handy for some kinds of occupations.

2045: What are the main obstacles? (I do not mean a complete transplantation of the brain into an artificial body, but at least remote control of the body).

A.A.F.: I don’t see any fundamental constraints on the solution to the problem, except that the problem is rather complicated and requires good collaboration of specialists in different disciplines.

2045: As we know, the brain controls many bodily activities, so what would happen to it if we ‘connected’ it to an artificial body? How complex should an artificial body be so the brain doesn’t notice the replacement?

A.A.F.:The question is not whether a brain-controlled device could be too simple, the question is that it shouldn’t be too complex. The problem here is not the lack of resources of the brain, but the lack of ways for a BCI device to understand and classify its intentions.

Now brain-computer interfaces are able to control only quite primitive devices, which have not many degrees of freedom. An artificial arm, for example, is controlled by commands such as, “Grab”, “Release”, “Clench the fist”, “Unclench the fist”, etc. But the movement has been programmed in advance, and when a manipulator receives a command, “Grab”, it moves along a predetermined trajectory and his fingers grab an object located in a predetermined place. “Bring to the mouth” command makes the prosthetic bring a glass to the point in space where the mouth is supposed to be, and the patient must bring his mouth to this location.

How many various human intentions can be recognized by an BCI and how many degrees of freedom can be controlled by one - it is a significant question which is now being solved by BCI researchers.

2045: Is an artificial body something internal or external to the brain?

A.A.F.: A brain is an organ aimed at meeting the needs of the body. These needs are not limited to renewing the resources of the body, they also include such needs as the need for social approval and the need for a place in a social group, etc. A brain regulates current needs and prioritizes the most important ones. It serves its body, therefore, if a body an a brain are separated in space, a brain will continue functioning only if he perceives the body as belonging to him.

2045: There are many processes in our body we rarely pay attention to: breathing, pulse, digestion, temperature regulation. Will a human be able to operate all these processes manually?

A.A.F.: Indeed, a human cannot consciously control individual muscles. He consciously sets a goal of motion, and subconsciously transfers this goal into muscle activity. However, during biofeedback experiments a goal of controlling a single muscle, or even a single neuron, was achieved. But he cannot realize what processes were happening in the brain. He learns from his mistakes by consulting the bio-feedback device, and eventually learns to reproduce the very brain activity which leads to activating the single muscle or the single neuron without knowing the precise nature of this activity.

In the case of brain-computer interface, bio-feedback is involved, too. Practicing to control a BCI, a man learns to suppress noises not related to the interface, while the activity related to the problem becomes more stable. Thus controlling external devices is similar to controlling one's own body in a way that no “manual” control of different degrees of freedom is needed.

2045: Human-computer interface is such an admirable idea! Is it possible to establish a communication between different people via this kind of technology?

A.A.F.: I don't see any problems with that. One man would show his intention on a computer screen via his BCI, an other would give him response via his BCI. It is possible to develop a device which would code an intention into not only a visual signal, but into an audio or tactile one. Eventually the intention will be coded into a signal which goes directly to the brain, making a man feel tingling or see auras, I suppose. The problem here is purely technical.

2045: Speaking of thought-controlled bionic prosthesis, can this prosthesis differ from a man’s hand? What difficulties are there in controlling this hand, or controlling a whole body consisting of such nonstandard items?

A.A.F.: There are no fundamental difficulties in assimilating additional hands and legs. The experience of using different tools speaks in favor of this. A skilled worker perceives his tool as a part of his body; there is no principal difference between using tools and using one’s own body.

But all the sensory organs should be located in the same body. Sensorimotor coordination means that the brain coordinates all the information from various sensory systems, and that’s why we perceive the world as being external and stable, not as a chaotic series of changes in receptor stimulation.

2045: If memory and consciousness are no more than products of the brain, is it possible to replace them too? Is it feasible to reconstruct consciousness into any external carrier? I mean total reconstruction, preserving the identity to fine details, like, say, political views?

A.A.F.: Consciousness, memory, and thinking are processes. They may be reconstructed, but it is incorrect to discuss any prosthetic appliances for them. As these processes serve a body, they may be reproduced only by the brain which does have a body. Either the brain, or the body may be artificial, though. I think rewriting one’s memory to an external device, like a computer disc, is not possible because of the very large differences in their organisation. But I imagine writing memory to a neural prosthesis that imitates the structure and the functions of a real neuronal network and is put into a brain to replace a damaged part. After integrating with the undamaged parts the prosthesis gets involved in a distributed memory system. Then a second prosthesis is put into the brain, etc. In this case, sequential addition of new prosthesis would occur naturally in a similar way as undamaged parts of the brain assume the functions damaged ones can no longer perform. It seems to me, that with this kind of memory transfer one’s memory and self-identification will be fully preserved.

2045: So, there are no insurmountable obstacles for creating a remotely controlled body at first, and then for life expectancy to increase several times by using a “living brain - artificial body” combination. It‘s overwhelming. Could you please tell us the scale of the social, financial and scientific requirements of realizing the project?

A.A.F.: I think, the scale and multidisciplinary approach of the project can be compared to those of space flight and nuclear power. By the way, I’ve recently read that China was planning to build a National Brain Research and Modeling Agency, similar to the American NASA.

2045: We’ve seen successive generations of high-tech cell phones replace each other with breakneck speed, while their capabilities dramatically increase. Does something similar happen in the field of neural interfaces?

A.A.F.: I expect wide implementation of this technology in the game industry, and in recovery for stroke patients, in the very near future. If it happens, incentives for further development of BCI will appear.

2045: Do people really need artificial bodies when sometimes it is hard to feed the natural ones?

A.A.F.: BCI research doesn’t intersect with this problem in any way. Moreover, the creation of an understandable and controllable project at the state level will help to improve the overall Research and Development level in the country, as happened during the nuclear and spacer projects. As I said before, the social utility of BCI research is related to its clinical usage, while commercial interest is related to the game industry. A stroke or even more serious disease may happen to anyone, so we should be interested in creating the treatments. Furthermore, we are all mortal, and most of us would not mind extending our lifespans. If the project we are talking about will contribute to that, it is of high importance, despite the probable ethical issues. I think none of the existing life extension projects are free from risk.