Hack the brain. When will scientists be able to read our minds?

Science, technology and technology news

Hack the brain. When will scientists be able to read our minds?

Hack the brain. When will scientists be able to read our minds?

For the first time in history, American researchers were able to “read the thoughts” of a paralyzed person. They did this with a neuroimplant that converts brain waves into meaningful words. The technology was tested on a patient who suffered a stroke that made him speechless 15 years ago. Now he can use about fifty words just by thinking about them.

Recognizes up to 18 words per minute

Every year, thousands of people lose their ability to speak due to accidents or some kind of disease, the same stroke. Scientists have been trying to help them for a long time: they have focused on technologies for reading the electrical activity of the brain. After all, a thought, whatever one may say, is just an electrical signal running through neurons. And if there is such an impulse, albeit very weak, it means that in the presence of supersensitive devices, it can be captured and transformed into something understandable. The efforts of neurophysiologists are aimed at creating such devices.

Scientists at the University of California, San Francisco have been developing a neuroprosthesis that can decode brain waves into words for more than 10 years. They placed electrodes in brain surgery patients who were fluent in speech to track electrical activity in areas that control their voice. So they analyzed patterns for word prediction and built up a basic vocabulary.

Then it was the turn of the paralyzed patient who lost the ability to speak. Since his cognitive functions remained unchanged, and he could partially move his head, neck and limbs, the researchers asked him to communicate with them using a pointer attached to his baseball cap – so he showed certain words and letters on the tablet screen. Electrodes were implanted on the surface of his brain, and the computer analyzed the brain impulses when the patient tried to pronounce certain words. In the end, a dictionary of 50 words (“water”, “family”, “good”, etc.) was typed, after which artificial intelligence was connected to the case. He learned to recognize brain activity, discern subtle patterns in data, and identify words in real time as soon as a person started thinking about them.

As the researchers report in their published article, the patient can now answer, for example, the questions “How are you?” and “Would you like to drink?” His mental responses appear on the screen as text. The system is capable of recognizing up to 18 words per minute, the accuracy is up to 75%, but the developers are confident that these indicators can be improved. Further, as they say, is a matter of technology.

“Speech usually transfers up to 150 or 200 words per minute. We hope we can move from brainwave recognition straight to speech rather than text. This will be a big advantage, says Edward Chan, senior study author and neurosurgeon at the University of California, San Francisco. “As far as we know, this is the first successful demonstration of direct word decoding based on the brain activity of a paralyzed person. It is good to think that we are at the very beginning of a new chapter in medicine designed to alleviate the suffering of patients who have lost the ability to speak. “

The development of technology depends on AI

One can only rejoice at the success of American neurophysiologists. But the question involuntarily comes to mind: is it realistic to develop the technology for recognizing the electrical activity of the brain to such a level that one can learn to “read the minds” of people? How does it even look like?

A couple of years ago, the Xinhua News Agency reported that Chinese engineers have developed a microcircuit that will read and decode human thoughts. Scientists at Tianjin University have created a device called the Brain Talker: it decodes the signals of neurons in the cerebral cortex, converting them into digital data. But the sensationalism of the news came down only to the size of the device – it was a chip. And the technology itself, called BCI (Brain-Computer Interface), has been known for a long time.

Initially, it was conceived to help immobilized people: sensors placed on the head (not necessarily implanted in the brain) should read their simple intentions, expressed in the same impulses of electrical activity, and translate them into concrete action. For example, move the cursor on the computer screen, turn on and off the TV, give commands to a wheelchair, operate a prosthesis, etc. It is clear that this is not yet mind reading and it is far from it, but the direction for the development of technology was chosen exactly like this – the same which neurophysiologists from California also followed.

Their colleagues around the world are developing similar ideas. It looks like future success will largely depend on artificial intelligence algorithms. Thus, a group of scientists from the Netherlands, Germany and the United States is conducting experiments in which the signals from the brain of the subjects are read by the neural network at the moment they utter a speech. Artificial intelligence analyzes the work of the brain, learns and selects the appropriate words and sentences. The accuracy is close to the same 75% achieved by the developers at the University of California, San Francisco.

How to tell my grandmother from Marilyn Monroe?

The main problem with recognizing brain impulses is that they contain a lot of background noise. It is extremely difficult to isolate a signal from a specific group of neurons in it. And to interpret it as a human thought and correctly formulate this very thought – the task, it seems, is generally unrealistic.

Scientists argue whether it can be solved in principle. And the overwhelming majority is sure: no, it cannot be solved. It is impossible to read thoughts with instrumental methods.

“Even if we find a connection between a group of neurons and a certain visual image (even such terms as“ my grandmother’s neuron ”or“ Marilyn Monroe's neuron ”have appeared), and then we learn to recognize this connection, it will still not be mind reading. After all, mental representations of a person in terms of content are not as simple as they seem at first glance. And the signals that we have learned to “pull out” from the brain are not thoughts at all, but just our intentions, ”explains the psychophysiologist, Doctor of Biological Sciences, Professor of Moscow State University Alexander Kaplan. – There is a second circumstance that prevents instrumental reading of thoughts when directly connected to the brain. Thought is the cumulative result of the activity of many parts of the brain. No matter how many electrodes we take, there will still be few of them. This is an extremely poor connection to the brain. Hundreds of electrodes are nothing compared to the astronomical complexity of the brain, which contains nearly 100 billion neurons and a million billion connections between them. Even 100,000 electrodes are clearly not enough. In addition, each pair of nerve cells has its own communication code, which still needs to be unraveled. “

The scientist believes that the task of reading a person's thoughts using instrumental methods is tantamount to the task of fully understanding the brain. And science is unlikely to ever be able to solve it. The common phrase that the human brain is the most complex object in the Universe is most likely not just a beautiful expression, but a statement of fact.

This is why he is amazing. It weighs only one and a half kilograms. But it contains a riddle that humanity is unable to solve.

Leave a Reply