How we reviewed Neuralink
Own comment on technology Neuralink we published in October 2019. Compared to last year’s presentation, the appearance of the device has changed – now it is a miniature round tablet the size of a coin and a tongue made of a bundle of electrodes, which, by the way, have become smaller. Initially, 3072 electrodes were announced, now there are 1024 left.
It can be assumed that the change in the number of electrodes is associated with the tasks of the device. The device aims to recognize specific brain activity. The presentation showed how it is used to read physical activity. For this task, a chip must be implanted into the motor cortex. This is a small area and 1000 electrodes are likely sufficient.
If we want to solve another problem – to stimulate the visual or auditory centers, then it is necessary to implant the chip in other areas of the brain. Moreover, to control complex cognitive activity, the electrodes must cover multiple areas at the same time. This is not a question of the number of electrodes, but of the safety of such an operation.
There are also new features. When asked by journalists whether it would be possible to control Tesla using Neuralink, Max answered positively, adding that all senses: sight, hearing, touch are electrical signals that are sent by neurons to the brain. Given the above, it can be assumed that the technology is applicable to control smart home and the Internet of things. The presentation also said that the chip can broadcast music and work with various devices via a Bluetooth connection, and a mobile application will be developed to control Neuralink. Musk likened the new prototype to Fitbit’s fitness bracelets with only small wires in the skull.
Let’s take a look at why the technology is new and why Neuralink could be a breakthrough in the treatment of cognitive impairment. To do this, we abstract from marketing taglines about chip-based control in StarCraft and conversations about chipping people.
Device prototype a year ago and now
Why the Mask project is promising
The idea behind Neuralink is based on a computer brain interface. The term “brain-computer interface” (BCI) appeared in the early 1970s, and the first attempts to study neural activity in monkeys were carried out as early as the 1960s. Today, work in this direction is promising for rehabilitation in case of motor dysfunction.
Neuralink enables the next generation of invasive techniques. The device contains up to 1024 electrodes, distributed over dozens of threads, with the help of which it is connected to the brain. To overcome the surgical limitation, the developers have created a neurosurgical robot that injects up to six threads per minute with micrometer precision.
The technology can serve as a prototype for an invasive neurointerface for clinical applications. Multi-electrode neurointerfaces can become the basis for new technologies and medical solutions for paralyzed people. The development of technology will make it possible to interact with the external environment without restrictions through integration into a smart home and the Internet of Things.
Neuralink has no analogues in terms of the number of registered channels. Existing BCIs, which use invasive recordings from several dozen neurons, already allow monkeys and humans to control the movements of the manipulator with the power of thought. The journal Nature published works that demonstrate how a monkey is eating robotic arm and how completely paralyzed patients grasp and move objects with manipulator…
Device strands interact with the brain
BCIs are promising for detecting hidden information about the work of the brain, which cannot be obtained using conventional communication channels. The use of non-invasive BCIs is limited by the small number of recognizable commands. This limitation arises from the noise and non-stationarity of non-invasive EEG or NIR spectroscopy recordings.
In this regard, invasive electrodes are more resistant to interference and artifacts, and allow obtaining high-quality records of neural activity. However, invasive recording requires more electrodes to cover the distributed areas of the brain. With the help of Neuralink methods, this problem can be solved.
How to Invasively Predict Epilepsy Seizures
For us, the Mask’s technology is interesting, including from the point of view of our own developments. At the Laboratory of Neuroscience and Cognitive Technologies, we are working on a project to prevent epileptic seizures.
We have developed a BCI that, using three electrodes implanted in the rat’s brain, can predict epileptic seizures with an accuracy of 90%. However, there are problems associated with the large number of false predictions. When it comes to a seizure prevention system using electrical stimulation, false predictions lead to a lot of unnecessary stimulation. Our interface was able to minimize the number of false predictions, but the seizure prediction accuracy dropped to 50%.