Facebook has published details of its brain-computer interface experiments, with some promising results. The system is a brain-reading computer developed by scientists at the University of California, San Francisco backed by Facebook Reality Labs. The paper was published in Nature Communications and details the team's success in reading brain impulses associated with certain words that the test subjects were about to utter.
There already exist brain-computer interfaces that people can use to type, but in order to do so, they need to spell out each letter individually on a virtual keyboard. The team at USCF demonstrated that their system could extract the answer to pre-determined questions based on the subject’s thoughts.
The experiment was performed on subjects who were on their way to epilepsy surgery and had invasive implants placed in various parts of their brain. Then the system was trained on the 24 pre-determined questions and its possible answers. Once the training was complete, the subjects would be presented with a question, to which, they would answer out aloud. The system would be able to detect their answers based on its training and lead author and UCSF neurosurgery professor Edward Chang said that the accuracy of the system was between 61 to 76 percent, far higher than the current 7 to 20 percent of existing such brain-computer interfaces.
While the study has promising results, it is far from what Facebook has been promising the world; a non-invasive brain-computer interface. The current paper is based on a system that requires highly invasive implants, and more importantly, is functional in a very limited environment where it has already been trained on all the questions and their potential answers. Lastly, the answers had to be spoken in order for the system to read the brain waves and deliver the result, which isn’t exactly “thinking and typing.”