Hearing the Internet, Speaking Without Sound: A Delve Into Wearable Neural Interfaces
“What’s the largest city in Bulgaria, and what’s the population?”
This isn’t the typical 60 minutes episode — Scott Pelley, paper in hand, sits across Arnav Kapur, researcher at MIT’s Media Lab, who’s wearing a white headset around his ear.
“Sofia, 1.21 million”
Kapur Googled the answer — in his head. Pelley scoffs, and confirms, just to be sure:
“You have the entire internet in your head”
“That’s the idea.”
Kapur did it with AlterEgo, a non-invasive, wearable, silent speech system for natural language communication with machines, artificial intelligence assistants, services, and other people — all without opening your mouth. You just need to think it.
MIT’s Media Lab says the main focus of the device is to help people with speech disorders, such as those with ALS or multiple sclerosis. The device captures peripheral signals, which are in the somatic system (nerves around the body) through internal speech movements, instead of capturing them directly from the brain. It’s thus somewhat obvious how it might help people with neurological disorders that directly impact their brain though not the rest of their nervous system.
That said, the primary concern behind AlterEgo isn’t really the device itself, or the technology that poses a threat — it’s rather the way that it might be applied broadly in various contexts. It’s unlikely that AlterEgo, indirectly commodified with epithets like “wearable”, “seamless”, and “natural”, won’t be applied in different contexts.
How it Works
Before we get to the broader implications, here’s a rundown of how AlterEgo works:
It starts off by picking up the user’s internal vocalizations, which are neuromuscular signals from the back of the palate and tongue.
Internal vocalizations are essentially when you voice words you read inside your mind. Think back to doing this in elementary school, when you initially would read new words aloud, but after your elementary teacher told you that was obnoxious, you resorted to voicing them internally and silently.
Over time, we unlearn this “silent speaking” in order to read faster. However, AlterEgo banks on these silent, internal vocalizations. Doing so creates signals from the brain to the relevant muscles, which are processed by AlterEgo.
The device catches these neural signals, enabling the user to transmit streams of information to and from a computing device or another person, without unplugging the user from their environment. Hardware and software is used to process these electrophysiological signals, including a modular neural network based pipeline that has been trained to detect and recognize words that are silently spoken by speakers. It will then process the user commands and queries to generate responses — this is when you ask, thinking the words aloud in your head, what the largest city in Bulgaria is and its population.
Lastly, AlterEgo will then give the user the auditory output through bone conduction headphones, which transmit sound waves through the skull rather than the eardrum — allowing the user to hear the input, while their natural listening continues uninterrupted. This is the part when you hear Sofia, 1.21 million.
Neurological Implications of Wearable Devices
According to MIT’s Media Lab, AlterEgo uses the power of computing to interface with devices without having to unplug from their environment or unplug from real-world interactions. It puts the power of computing within a user’s self as opposed to with their fingertips.
Beyond uses for those with neurological disorders, AlterEgo might allow for faster Google searches, ease of telecommunication from leveraging the bandwidth of vocal speech, with the addition of fidelity that silent speech provides, as well as allowing for a storage of memory in the device.
Smartphones require a shift in attention to consult a device, whereas AlterEgo is less socially disruptive and almost acts as a readily available extension of the self. Though the creators of AlterEgo claim this allows users to be more closely connected to the real world, there may be potential for detrimental neurological compensations made for the device.
Consider this: a 44-year-old man from France reported weakness in his leg, after which a scan revealed he’s missing 90% of his brain due to hydrocephalus. The caveat is that he was living a normal life — he has a family, works a job, and is socially apt.
His brain was able to work around missing structures in radical ways to allow for normal functioning. However, with the introduction of a new neural structure, that acts as an extension of memory, there’s no way of delineating the extent to which the brain itself will disinhibit its normal function from relying on the device. When the device comes off, it may become harder for cognitive circuits to reorient themselves to their biological structures. Key cognitive functions, such as active recall, are compromised when the entire repository of human knowledge is a search away — the brain prefers the path of least resistance, and this is likely the route it will take in a normally functioning brain.
The Long-Term Vision
A novel paper published in August this year described how scientists reconstructed a Pink Floyd song being listened to by participants using brain decoding analysis not dissimilar to AlterEgo. Dr. Daniel Toker, neuroscientist at UCLA, summarized the article in an Instagram reel with nearly 4 million views. The top comment?
“Remember kids: eventually, this too will be weaponized.”
Cynicism aside, wearable neural interfaces could compromise autonomy and privacy: cyberattacks might be against the mind if widespread commercialization occurs, criminal sentence aggravation may be justified by decoded data supposedly unearthing lies, and companies may use query and output data to personalize advertisements. The list goes on.
Wearable neural interfaces do seem to reduce the so-called disruption of picking up a phone and typing out the words, though the neurological costs are nebulous. There’s no saying who or what will leverage the technology for monetary or political benefit.
That’s not to say the technology should be abandoned. Strict regulation is warranted, especially considering the extent to which it benefits vulnerable populations with neurological deficits. Ultimately, the priority needs to start with the principles of autonomy and privacy, and the development of technology should proceed from that foundation.
This article was written for QMIND — Canada’s largest undergraduate community for leaders in disruptive technology.
