Forget typing status updates on social media as you sit on the toilet, cross the street, or wait in line at the coffee shop. Forget clumsy thumbs, awkward autocorrect and neck cramps from staring down at your phone.
Facebook wants to make it possible for you to think your social media posts.
Harnessing a rapidly developing class of technology called “brain-computer interface,” or BCI, the global social media giant hopes one day to offer users a wireless brain modem that can translate and encode your brain activity and interface with external devices.
Facebook isn’t alone in wanting to broadcast user’s thoughts. Paypal billionaire Elon Musk in July announced his own plan to develop a commercial brain modem.
Incredibly, the basic technology already exists. It’s just not very reliable. Worse, to work with a reasonable degree of accuracy it requires that a surgeon drill holes in your head.
Still, it’s perhaps a sign of things to come that Facebook and Musk seriously are spending real money developing a brain modem. It’s possible, even likely, that in coming years or decades we’ll interact with social media merely by thinking about it.
Facebook first announced its BCI initiative in 2017. The goal, the company stated, is “to build a non-invasive, wearable device that lets people type by simply imagining themselves talking.”
To that end, Facebook signed up as a major donor to one of the world’s leading neuroscience laboratories. The University of California San Francisco’s Chang Lab, headed up by neurosurgeon Eddie Chang, helps people who’ve lost the power of speech—from stroke, spinal injury or a neurodegenerative disease such as ALS—to communicate again.
Now Chang and his lab have achieved what appears to be an important breakthrough. One that could nudge Facebook a little closer to its goal of mind-controlled social media.
In a July 30 article in the scientific journal Nature Communications, Chang and co-authors David Moses, Matthew Leonard and Joseph Makin revealed they had succeeded in translating brain activity into text with up to 76-percent accuracy.
“The researchers recorded activity in areas of the brain that tell the muscles of the lips, tongue, jaw and voice box to make specific sounds during speech—e.g., ‘ba’ versus ‘gu’—which let them figure out which words the volunteers were probably saying,” Nicholas Weiler, a UCSF spokesperson, told The Daily Beast via email.
If you can track the electrical impulses in the brain that correlate with certain words, you could in theory encode those impulses and transmit the data via modem to an external device. It’s thought-to-text. Or, as Jacob Robinson, a Rice University neuroscientist, described it to The Daily Beast: “A fancy voice assistant that would work by imagining a command, but not actually speaking it.”
Facebook celebrated Chang’s July breakthrough as a step toward the company’s own goal of "a real-time decoding speed of 100 words per minute with a 1,000-word vocabulary and word error rate of less than 17 percent.”
The tech, even in its crude form, soon could prove useful for people who’ve lost the power of speech and whose willingness to take risks is high. But it’s not nearly ready for widespread use by folks who can still talk and type, but prefer not to as they post on social media.
Facebook itself in a July statement acknowledged “how far we have left to go.” The social media company declined to answer specific questions for this story.
Daniel Palanker, a Stanford neuroscientist, told The Daily Beast that, given the state of the art in BCI tech, Facebook’s 100-words-a-minute goal is “wishful thinking.”
For one, the hardware in its current form is invasive. Chang’s volunteer test subjects all had electrodes temporarily implanted in their brains as part of treatment for epilepsy, Weiler explained. The researchers hooked up their own sensors to the subjects’ existing medical implants.
To avoid risky surgeries and to streamline regulatory approvals, a commercial version of the brain modem probably would need to be non-invasive: a patch or a headband or some other wearable device. Indeed, the word “non-invasive” is all over Facebook’s various statements regarding its BCI ambitions. Facebook does not want to drill holes in your head.
The problem is, today’s sensors lack the sophistication to read brain waves through bone. “Facebook is talking about recordings from outside the head,” Andrew Schwartz, a University of Pittsburgh neuroscientist, told The Daily Beast. “Those signals have a very low information content and it is highly unlikely, from a basic science perspective, that they can get anything like the performance they are after with this type of recording.”
Chang’s own lab seems to recognize these limits. “Based on our current knowledge, decoding speech is only feasible with electrodes implanted under the skull,” Weiler explained.
There’s another, more fundamental obstacle to translating thoughts into social-media posts. It’s one thing to map the brain activity associated with deliberate efforts to speak. It’s another to capture thoughts that aren’t associated with speech.
Weiler told The Daily Beast Chang and his team identified brainwave patterns for “a couple dozen” words and phrases. The researchers instructed their test subjects, in response to questions, to think about speaking only those words or phrases. Someone using Chang’s current system more or less would have to silently mouth the words they want to include in their mind-controlled social-media posts.
It would be entirely another thing for someone to idly transmit a vague impression—“I’m happy” or “I’m sad” or “feeling cute”—to their newsfeed. “At this point, we wouldn't know where to record from to decode inner thoughts much less how to decode the brain signals to recover those thoughts,” Robinson explained.
Still, Robinson for one said he’s optimistic. The tech will improve. People who’ve lost the power of speech could benefit first. After that, the general public could start thinking thoughts directly into their social networks.
And it’s a good thing that Facebook is so open about this possible future so far in advance of it becoming a reality, Robinson said. “Facebook has reached out to the neuroengineering community to work on neuroethical issues even before they reach the point where they have neurotechnology products.”