Issue 709|2022/01/03
Imagine controlling the outside world without any words, with only thoughts, and exchanging ideas with only eye contact with a companion — and in the imagination of Brain Computer Interfaces, these scenarios that exist only in science fiction will become a reality.
In the 1970s, Jacques Vidal, a computer scientist at the University of California, Los Angeles, first proposed the concept of brain-computer interfaces. Since then, brain-computer interface technology has evolved, from using electrode caps to drive the cursor with the mind, to implanting electrodes into the brains of stroke patients for the first time, to successfully enabling paralyzed people to perform more complex movements, scientists have crossed barriers step by step, achieving significant progress.
Across the globe, business giants such as Meta (formerly Facebook), Google, Amazon, and Neuralink, the founder of Tesla, Elon Musk, are actively laying out the field of brain-computer interfaces. So how far has the brain-computer interface evolved? How are brain signals detected and interpreted? How far is "mind control" from ordinary people?
The "Daily Economic News" (hereinafter referred to as the NBD) reporter interviewed two "big bulls" in the field of brain-computer interface - José Millán, academician of the International Association of Electrical and Electronics Engineers and professor of electrical and computer engineering at the University of Texas at Austin, and He Bin, professor of biomedical engineering at Carnegie Mellon University.
Bin He leads several national institutes of health and National Science Foundation-funded training programs, and his lab first demonstrated in 2013 that humans use non-invasive brain-computer interfaces to manipulate drones and control robotic arms. Around the design of brain-controlled robots, Jose Milan has made several pioneering contributions in the field of brain-computer interfaces based on EEG signals, and has been named a fellow of the International Society of Electrical and Electronics Engineers and an academician of the International School of Medical and Biological Engineering.
Decoding the Brain's "Intent"
NBD: Can you introduce the technology of brain-computer interface to the public? How are the signals transmitted by the brain detected and interpreted by machines? What is the current stage of research on this technology?
He Bin: Brain-computer interface is a neurotechnology that integrates the brain with a computer or external device, and its main application is to decode brain signals and use decoded signals to control devices. Brain "intent" is encoded in neural circuits that produce small electrical signals that travel through brain tissue and are recorded by sensors, such as electrodes implanted inside the brain or on the scalp. Brain-computer interface research is still in the early stages of its development, but it has shown great application prospects.
Jose Milan: Brain-computer interfaces are a technique that can detect brain activity, divided into invasive (implanting electrodes in the brain) and non-invasive (not requiring anything implanted in the brain). Non-invasive brain-computer interfaces detect brain activity from the outside, one way is to use an electrocardiogram, which simply attaches electrodes to the person's head, and the other way is to use FMRI (functional magnetic resonance imaging) technology.
The obtained brain signals are passed to the computer, which is then converted by signal processing. We will apply machine learning and artificial intelligence techniques to build a model of how the brain decodes different mental instructions, decoding brain signals so that the person's intentions are transmitted to the computer, which in turn sends it to the next level, such as a wheelchair, so that it can be moved anywhere.
NBD: What is the difference between non-invasive and invasive brain-computer interfaces? Is the current research focused on invasive or non-invasive?
Bin He: The invasive brain-computer interface can record brain signals in high fidelity and signal-to-noise ratio by implanting electrodes in the brain close to neural circuits. This led to many important neuroscience discoveries, but at a cost that was invasive and therefore rather limited to most people.
Non-invasive brain-computer interfaces can be easily applied to humans without creating ethical or security concerns. However, its limitation is that the signal quality is low and the decoding is difficult. At present, both invasive and non-invasive brain-computer interface studies have received widespread attention.
Jose Milan: The two brain-computer interfaces have different advantages and disadvantages. Invasive brain-computer interfaces can be very close to the source of electrical activity, but electrodes need to be surgically implanted to end up with high-resolution images of a small fraction of the brain.
Through non-invasive brain-computer interfaces, you can gain insight into the overall picture of brain activity. When we are planning and performing physical movements, many areas of the brain need to coordinate to perform those movements. Thus, although there is no high-resolution view of the activity of individual neurons, it is possible to reconstruct the activity (signals) of the brain. In addition, non-invasive brain-computer interfaces require no surgery, are cheaper, and are easier for ordinary people to use.
To some extent, invasive brain-computer interfaces can detect brain signals more precisely, but only a small fraction of them can be seen. If a larger field of view is required, it cannot provide more additional sources of information.
Can emotions and memories be decoded?
NBD: How are brain-computer interfaces (especially invasive) assessed as a person's health risks during the course of the study?
Bin He: The current invasive brain-computer interface requires electrodes to be implanted into the brain, which can cause damage to brain tissue. Human trials of invasive brain-computer interfaces are primarily targeted at patients with severe paralysis or other medical conditions. There is currently no data on the risks to human health using invasive brain-computer interfaces, and risk factors depend largely on the course of surgery.
Jose Milan: According to my colleagues who work on invasive brain-computer interfaces, they have not yet observed that patients lose any of their previous abilities as a result. While some local damage may occur, it is not enough to cause additional loss of capacity, so this risk appears to be very limited. For non-invasive brain-computer interfaces, we did not observe any risks.
NBD: Scientists have tried a lot about decoding motor functions, is it possible to decode other functions such as emotions, pain, and memory through brain-computer interfaces in the future?
Bin He: The brain-computer interface can certainly help decode brain states and functions beyond motor function. For example, my lab decodes pain levels in human subjects through non-invasive EEG, but these studies are currently in a relatively early stage. In my opinion, the most important challenge is how to decode the function and state of the brain under unfettered conditions, not just motor function.
Jose Milan: Advanced functions of cognitive decision-making and memory processes are encoded in our cerebral cortex (the outermost part of the brain), but human emotions are mostly encoded in the subcortical region, which is difficult for brain-computer interfaces to access because they are located below the cortex. As a result, it is relatively more difficult to decode emotions, but can be judged by secondary effects of the cerebral cortex.
Remodeling "neuroplasticity" restores impaired function
Image source: Courtesy of He Bin's team
NBD: What is the current research focus of your lab?
Bin He: My lab studies systems neural engineering and develops new sensing, imaging, and stimulation techniques to study the central nervous system. Currently, we are focused on developing novel non-invasive neurotechnology that can image brain function and dysfunction, regulate the brain, and connect the brain to external devices to enable "brain-controlled intelligent systems."
Jose Milan: There are two main points of focus at the moment. We have developed reliable brain-computer interface technology, and how to make it clinically available to benefit people with severe neurological disorders is one of the focuses. (Brain-computer interface) not only enables paralyzed people to drive their own wheelchairs, but also helps them regain their functions. Our ongoing clinical trials show that patients with severe hand disabilities can regain hand function.
The second point is more fundamental, the brain-computer interface can enable people to learn to regulate brain activity in different ways in order to send better instructions, and we are accelerating the learning process of brain-computer interface users.
NBD: How do brain-computer interfaces restore function to patients?
Jose Milan: When the brain is damaged, it actually has the ability to change, which we call "neuroplasticity." This plasticity manifests itself every day of your life, such as when you learn or memorize new knowledge. We are using brain-computer interfaces to reshape these unknown plasticity processes so that brain regions that have lost their coding function can regain function.
How far is it from universality?
NBD: You mentioned in a previous interview that there is a new direction in invasive brain-computer interfaces, namely bidirectional brain-computer interfaces. What is the current state of research in this field? Are there any other emerging research directions?
He Bin: The two-way brain-computer interface is still under study. In addition, an important research direction is the introduction of artificial intelligence (AI) and machine learning into brain-computer interface research, and the artificial intelligence revolution may significantly advance the research of brain-computer interface.
My lab recently demonstrated that using deep learning, we were able to improve the performance of non-invasive EEG-based brain-computer interfaces in 62 healthy subjects.
NBD: At present, there have been successful cases of implanting brain-computer interfaces into the human body at home and abroad. How far away is this technology from becoming widespread?
He Bin: There are indeed successful examples, but the number of subjects with invasive brain-computer interfaces is still very small. It's a promising technology, but it's still a long way from being widely used in patients, not to mention healthy people. Both invasive and non-invasive brain-computer interfaces require a lot of effort to achieve a wide range of clinical applications, including the development of novel and reliable hardware and decoding methods to reliably and accurately read neural code.
Jose Milan: On the one hand, there are some limitations in the technology of brain-computer interfaces. We know that there is a lot of instability when recording brain signals, and how to ensure the stability of signals is still being studied. For patients, the high cost is the biggest limitation. This is also the goal of all research work – to show that brain-computer interfaces have enough advantages to strive for social health insurance coverage.
The future may enter the field of smart cars
NBD: In the future, how will brain-computer interface technology change people's daily lives?
Bin He: Brain-computer interfaces can help paralyzed patients regain their mobility and help other patients with various diseases. It can also help the general population, just like smartphones. It is for this reason that I have been working on non-invasive brain-computer interfaces, and only non-invasive brain-computer interface techniques can benefit the general population.
Jose Milan: Brain-computer interfaces can improve basic (behavioral) abilities in groups with disabilities, and can also help people with cognitive impairments (e.g., memory impairment, Alzheimer's disease, etc.) improve cognitive abilities.
In addition to this, creating usable brain-computer interface programs for healthy people is a big market in the future. Obviously, the demand of healthy people is not to do what can be done through brain-computer interfaces, but to hope that it will help us do what we can't do or predict what we are about to do. Current artificial intelligence does not take into account the cognitive state of the subject, if we can integrate the subject's cognitive system into artificial intelligence, we can create additional value through brain-computer interfaces. While using brain-computer interfaces, your brain also goes through a process of continuous learning.
NBD: "Brain-computer interface + medical" is an application direction that we now generally mention, in addition, what other application scenarios may be there?
Jose Milan: There is also a lot of potential in the field of smart cars, and one of the directions we are working on is to use brain-computer interfaces to serve smart car drivers, so that smart cars have sensors and "autonomy" to achieve autonomous driving. One of the biggest challenges is how to make the car drive in a human way, to do things that fit human ideas. We have been conducting studies for 8 years and the results show that this is feasible.
NBD: Looking around the world, many business giants such as Facebook, Google, Amazon, Neuralink, etc. are actively laying out the brain-computer interface field. How do you see the commercialization of brain-computer interfaces?
He Bin: Big companies are sensitive to future technologies, and brain-computer interfaces are one of those technologies, so it's not surprising. Without industrialization, it is impossible to translate basic research into greater social benefits. More and more commercial companies are interested in this, indicating the advantages and great prospects of brain-computer interface research, which will advance the research of brain-computer interfaces, which in turn will also help to bring it into society and the general population.
Jose Milan: I'm glad that big companies are interested in this area of research, which confirms the value of these studies to help human beings. But we need to note that we should not blindly expect too much from this. Imagine how many years artificial intelligence has been around? 70 years. But it wasn't until recent years that we saw its immense success.
For businesses, this is a business opportunity. But when one day they stop investing in this research, what will funders and the public think? Is it because brain-computer interfaces are no longer valuable? I don't think so. So I think we should be cautious about that.
Reporter | Wenqiao
Editor| Lan Suying, Sun Zhicheng, Du Bo
Visual | Cai Peijun
Typography | Lan Suying Wang Shujie
| This article is copyrighted by "Every Toutiao" |
Reproduction, excerpting, reproduction and mirroring without permission are prohibited
If you need to reprint, please apply to the background of this official account and obtain authorization