The next step involved a bit of machine learning: "With the recordings in hand, Chang and his team built computer models that learned to match particular patterns of brain activity to the questions the patients heard and the answers they spoke", reportsThe Guardian.
Researchers in the Facebook study - which was published on Tuesday in Nature Communications - said they were still only able to recognize a limited number of words and phrases.
Researchers at Stanford University, have already found a way to do so with paralyzed patients, but it requires surgery to implanted electrodes in the brain.
There's definitely nothing dystopian to see here; just one of those run-of-the-mill stories about how scientists have figured out a way to hook people up to wires, give them some cues, and then guess - with startling accuracy - what it is they're probably thinking. The predictions were shaped by prior context - so once the system determined which question subjects were hearing, it would narrow the set of likely answers. It wasn't 100% accurate, but it in some tests had a 76% accuracy.
"We know the technology better than anyone else, so we should start talking about this now with people in the community", said Emily Mugler, an engineer on the BCI team at Facebook Reality Labs.
There were 24 questions in total, two of which asked the patients how much pain they were in and which musical instrument they preferred.
The study essentially took three epilepsy patients who were about to have neurosurgery for their condition.
Back in 2017, Regina Gugan, the former head of the US Defence Advanced Research Projects Agency (DARPA) asked Facebook's developer conference: "What if you could type directly from your brain?"
Another reason why Facebook is eager to come up with technology that reads human brainwaves is to develop augmented reality glasses.
Meanwhile, Elon Musk's Neuralink company has applied to U.S. regulators for permission to start testing its own brain-hacking device on humans.
Facebook wants its wearable headset to be "non-invasive", so it has been investing in technologies that can monitor your brain activity from outside the skull using laser or fiber optics.
"Currently, patients with speech loss due to paralysis are limited to spelling words out very slowly using residual eye movements or muscle twitches to control a computer interface", said Edward Chang, a speech neuroscientist and one of the paper's authors. In the past, the technology has allowed people to feed themselves, and to fly a jet simulator, says Facebook Reality Labs research director Mark Chevillet. "And while measuring oxygenation may never allow us to decode imagined sentences, being able to recognise even a handful of imagined commands, like 'home, ' 'select, ' and 'delete, ' would provide entirely new ways of interacting with today's VR systems - and tomorrow's AR glasses".