Is Mind-Reading Technology Approaching?
How Brain-Machine Interfaces Are Developing
January 16, 2018, Bruno Jacobsen
No one can read minds. But the incredible progress we have made with technology may soon change this. Brain-machine interfaces are continuously improving and offer great potential. In essence, they could change how we interact with the world around us. Let's have a look at some developments that provide an extraordinary example of this.
Mind-Reading Technology : How Far From Reality Is It?
A lot of progress has been made recently in the area of brain-machine interfaces. Here's some of it.
1. Moving objects with your mind
In terms of manipulating objects, we have been able to play games, move mouse cursors and control robot arms with our minds. In 2006, a teenager suffering from epilepsy was able to play through an entire level of Space Invaders using just his thoughts. This was done with a technique called electrocarticography. In short, a grid atop the teenager's head recorded brain surface signals. By programming the Atari software to communicate with the grid, the ship could be controlled by certain thoughts.
More recently, in 2011, researchers used the same technique to move a computer cursor. Albeit somewhat differently. This time, the computer cursor was moved not by imagining a range of movement, but by thinking of a range of vowel sounds.
The difference between the two, according to researchers, is that the latter allows for a "discrete" type of control.
"(It's) for the same reason that you don't type a paper with a mouse - you have a keyboard with a number of discrete commands," they told BBC.
This allows people not only to move a mouse, but also to "click it." If you think of mobile phones, we are able to do an extraordinary number of things by moving a cursor and clicking. Same with computers. So these techniques could significantly improve the lives of people with impaired motor or communication skills, as several applications can be built on top of them.
Other advances in this area have included control of robot arms. In 2016, using the same technique, subjects have been able to pick up objects from a fixed location or move them from tables onto shelves. For the first activity, the success rate hovered around 80%. For the second, 70%.
We're still not there, but the progress has been remarkable.
2. Seeing and recreating pictures you imagine
Scientists have also developed an algorithm that can accurately reproduce images in the brain.
For a while already, we have been able to measure brain activity and identify what type of image people were thinking about. This is done by showing the brain several images and recording the activity. We can then ask someone to picture something and match the brain activity to certain images. But those images had to be "stock images" and the technique didn't allow for image recreation.
This time, they build a deep neural network (DNN), able to extract information from different layers of the brain's visual system. Using the translations in the DNN, the researchers can then begin building the images. By comparing the DNN activity with the templated DNN activity, images images can be recreated, and adjustments made pixel by pixel where the DNN activity and the templates differ.
In trying to analyse images completely imagined by people, shapes did better. While photos were not very successful, shapes were recreated with accuracy 83% of the time.
3. Reading thoughts and reproducing them through language
Finally, using an fMRI scanner and machine learning algorithms, scientists have also recently been able to predict what sentences people were thinking about.
They put together 240 complex sentences. Then, they had participants reading them, leaving one of the sentences out. By analysing the patterns activated in the 239 sentences read, they were able to predict the last one. They repeated this 240 times, one for each sentence. In the end, the technique was able to predict the missing sentence with 87% accuracy.
To improve on this, it seems like there will be some grunt work. Namely, they may need to categorize topics by the brain regions they activate. But if we are able to create a map of these categories, it could be possible to literally read someone's thoughts.
Such progress could help many people who are paralyzed or have lost the ability to speak by giving them the ability to seamlessly translate their thoughts.
The applications of mind-reading tech
These techniques have several applications. And there is a lot more research into brain-computer interfaces. However, these three alone show that we might be able to control robotized limbs with our minds. We could be able to recreate mental images and communicate through them. Or simply think of sentences and see them faithfully reproduced on screens or elsewhere.
This no doubt could improve the lives of those afflicted with some impairment. But it could also be used to enhance the lives of completely healthy people. And even though this is decades away, the progress is quick enough to warrant some serious debate about the ethical consequences and merits of such enhancement.