Still lot of questions to be answered about brain-computer interface devices

On Friday, Elon Musk livestreamed the unveiling of a working prototype of his newly-founded company Neurallink's much-anticipated brain-implant device, along with a pig which had a microchip implanted in its head.

By planting a coin-like terminal into the brain of the pig, they could wirelessly collect the neural signals from its brain and show the activeness of some neurons when people touched the pig on the nose.

Although the technology is still in its early stages, the company aims to develop the means to link human brains with machines as a prescription for medical ailments and disabilities such as memory loss, blindness and paralysis. According to Musk's previous interviews, such devices could, in principle, repair problems with the signaling functions of the brain and so improve eyesight and hearing, recover limb movements, even cure epilepsy.

As a scientist myself, I look forward to the day when the technology Musk describes can solve these problems. Yet every one with basic natural science training knows how many difficulties one has to overcome to turn "in principle" into practice.

Unlike other companies, Musk's team chose to implant the "coin" into the pig's brain, which we call an "invasive implant". In that process, there is risk of the pig's immune system attacking this foreign body, which might cause injury. Even if that harm could be under control, scar tissue may form around the wounded area, which makes it difficult to collect signals from neurons. This will be a major challenge.

The tiny implant has more than 3,000 electrodes attached to flexible threads that are capable of monitoring around 1,000 neurons. Musk's implant transmitted the signals from about 500 neurons in the pig's brain, but compared to 80 billion neurons in a human brain, this number is tiny. To cover the whole human brain will be very difficult because the electrodes and affiliated technology will have to be made a lot smaller.

Even if we can track every one of these 80 billion neurons, we are still far from understanding their collective activities, namely our minds. It is quite possible that there are new principles that only emerge on the level of many neurons. We are only in the early stage of understanding how the brain works. The huge progress of artificial intelligence in recent years is not based on mimicking human brains, but based on improvements in data computational capabilities, and deep-learning algorithms, all of which have little to do with brain science. Therefore, the hope of controlling the brain via controlling a few neurons seems overly optimistic.

There are many technological challenges to be overcome and ethical questions that will need answering before Neuralink can put its devices to the purposes intended.

－YUAN LANFENG, AN ASSOCIATE PROFESSOR AT THE UNIVERSITY OF SCIENCE AND TECHNOLOGY OF CHINA