A new machine learning algorithm, funded by the US military, would be able to isolate and decode patterns in brain signals to anticipate the movements of soldiers. It could therefore provide real-time feedback or recommend corrective actions before they even take place.
The discovery was relayed by the US military website, and is based on a study conducted by a team of researchers from the University of South Carolina’s Viterbi School Of Engineering, published in the prestigious scientific journal Nature.
Isolating neural signals
Before movement takes place, neural signals activate in a very specific area of the brain. It has long been difficult to identify areas of the brain stimulated by specific movement, with many actions taking place simultaneously. For example, if I move my finger, but I’m thirsty at the same time, a double brain signal will be emitted. Managing to isolate the signal that relates to finger movement alone posed huge challenges.
But the improvement of algorithms has led to great strides being achieved in this area in recent years. Through machine learning, by recognizing patterns of brain activity that repeat themselves, these algorithms are able to isolate the signals corresponding to the area of the brain specifically stimulated during movement. This has notably made it possible to develop brain-machine interfaces, or brain-computer interfaces, i.e. systems capable of triggering the movement of an object through thought, thanks to a computer which interprets brain signal activity.
These new interfaces have made it possible to make spectacular progress in the production of prostheses for the hands or legs for the disabled. A small brain implant, which translates their thoughts into actions, allows them to have much more precise movements of their fingers, for example (as in this video demonstration from the University of Michigan).
The brain-machine interface is also the concept on which Elon Musk’s company, Neuralink, is based, which wants to “connect computers directly to the human brain” to make us more efficient.
With a chip in his brain, Musk’s “super man” could think faster, increase his memorisation capacity, and make better decisions in real time.
Today, these models have made further progress and this is the whole point of the discovery made by the teams of Dr. Maryam Shanechi. Their new algorithm is even more efficient in the identification of brain signals and their interpretations.
The researchers tested it on brain data sets while performing various arm and eye movements. It greatly outperforms existing algorithms in identifying the distinctive neural patterns that drive these movements. The activation zones are therefore recognised with much more precision.
The possibilities behind this are staggering. Researchers can now predict the movement that will be performed just by looking at the activated areas of the brain on the monitor screen. A real tool capable of reading our thoughts… Or, at least at first, anticipating our movements.
One can very well imagine how the army or the police force could use this discovery. And it is no coincidence that the US military funded the project.
In a combat situation, the algorithm could give real-time correction indications on the accuracy of the soldier’s shots according to the intention detected (the area they are targeting), external parameters (distance, wind) and their condition (stress level, fatigue, etc.)
Secure and control weapons?
The tool could also be used by police to help law enforcement target areas of the body that can incapacitate the attacker if necessary without putting them in danger of death.
This is not the first time that technology has tackled the issue of securing weapons and their use.
But these smart guns have often been singled out by the very powerful NRA (National Rifle Association, the arms lobby in the United States).
The NY Times even carried out an investigation into the anger that such devices had generated at the time, because these safeguards would prevent a proper functioning of the gun. In fact, it is above all the fear of seeing more controls on firearms that motivates criticism.
It will be interesting to see what kind of reception the NRA and others will reserve for such algorithms which could, of course, prevent certain police blunders by offering shooting assistance, but also introduce much more control and monitoring of the use of weapons.