Carbon and Silicon Find an Interpreter
by Nicholas Gentry, B.S. | January 17, 2019
You awaken and roll out of bed. Your alarm stops as it receives signals from your brain indicating wakefulness. Entering the bathroom, you turn on the lights and smart mirror in front of you with your thoughts. As you brush your teeth and prepare for the day, the mirror feeds you data on the quality of your sleep, your general health, and schedule. You scroll through this information without touching any surface.
This scene is part of a future envisioned by the enthusiasts, entrepreneurs, and executives pioneering Brain-Computer Interfaces (BCIs). Despite the roadblocks present in bringing the technology to consumers, there is a thriving ecosystem of “neurotech” players traversing these obstacles and positioning themselves to capture market share as the industry begins to bloom.
The term BCI covers a spectrum of technologies united by one principle: the process of communicating directly from the brain to an external machine. These technologies can be broadly separated into ‘passive BCI’, where an individual’s unconscious brain signals are read and translated to a computer; and the more complex ‘active BCI’, where users can control software directly (for an engaging and extensive explanation on interfacing with machines, read Tim Urban’s piece here). Although fMRI and other biomedical imaging techniques can collect neurological data, by far the most common form of communication in contemporary BCIs is non-invasive electroencephalogram (EEG), where electrodes placed on the scalp pick up the macroscopic electrical signals generated by neuronal firing.
Large-scale neural activity is primarily patterned in waves, and various features of these waves are associated with certain cognitive phenomena. ‘Delta power,’ for example, is the activity within the 1-4 Hz frequency range and is strongly linked to sleep. The ‘p300’ is a well-characterized spike that appears during discriminatory decision making regarding a stimulus. Filtering and identifying these sorts of signals allows EEG-based BCI to eavesdrop on what’s going on in your head.
Such technology may call to mind a wholly hands-free future, but Graeme Moffat, Chief Scientist at the neurotech company Interaxon, believes full “brain control of a computer or machine will likely never be more than a tinkerer’s niche.” Navigating something as sophisticated as a computer with active BCI is extraordinarily complex and inefficient. “Everyone likes moving things with their minds, but as one engineering leader in the BCI community has said: after ten minutes people get tired of on/off and ask for the mouse and keyboard,” he says.
Instead, we can expect passive BCI and much simpler binary applications of active BCI- such as turning on and off devices- to dominate the emerging market. Nevertheless, this technology is still enormously powerful. Reliably interpreting even basic information directly from our minds has the potential to reform standard human-software interaction on a scale comparable to that witnessed during the smartphone revolution.
EEG is nothing new. The same basic functionality has been around for nearly a century. Only with the ongoing decreases in the size and cost of electronics has consumer-level manufacturing become practical. Even so, real-world applications for the technology, necessary to drive interest and further investment, have until recently been somewhat lacking.
Interaxon has had significant commercial success targeting one such application with their BCI device Muse, a simple, sleek headband which reads and provides visualization of brain signals during meditation. Users improve their practice by learning to recognize and enhance certain elements of their cognition, a process known as ‘neurofeedback’. Dreem, another successful BCI startup based in the Bay Area, sells a sleep wave-monitoring headband that aims to help people improve their sleep habits and alleviate their insomnia. Dreem and Muse, with combined unit sales in the hundreds of thousands, represent the vanguard of consumer neurotechnology targeted towards a general audience.
Other neurofeedback tools, operating off similar premises, hold promising implications for future therapies. The pathologies of some psychological disorders, such, as ADHD and depression, are associated with signal abnormalities in the brain that could be alleviated through training with a BCI.
Moffat claims these treatments are just on the horizon; “We’ve only just scratched the surface of EEG’s practical application in the measurement of neuroplasticity in health and disease. There are a lot of low-hanging fruit in the brain diagnostic and performance spaces that we can really make a big difference in with new analysis tools for EEG.” A patient with ADHD, for example, could self-monitor their attention-regulating EEG signals and, when paired with brain training software of the sort Akili Interactive is developing, use this information to optimize their performance on focusing tasks. The goal, currently in clinical testing, is that neurofeedback will help these patients learn better attentional control through an improved intuitive understanding of their own brain.
“As far as therapeutics, it’s about much more than just the BCI or the EEG device. The key elements are the tasks and exercises that someone is doing with the BCI, and the skills they’re learning,” Moffat explains. These tools will require much more patient work than a simple drug or surgery entails, but conversely is expected to involve much fewer side effects and durable results.
The most compelling demonstration of the principles of BCI neurofeedback is found in recent work with people suffering from paralysis, which has produced results resembling science fiction. Patients surgically implanted with electrodes can use on-screen keyboards, navigate mice, and even operate robotic arms, albeit after an extensive training period. These technologies are even being combined with electrical output signaling to allow users to regain rudimental control of their own limbs or exoskeletons.
Although exciting, these use cases are fairly narrow implementations of a technology that has potentially paradigm shifting implications. How might the greater population see such brain-computer communication used in their everyday life? Addressing this potential first requires tackling the problem of accessibility, one of the primary obstacles BCI manufacturers are currently navigating. Dreaming up potential real-life applications for ‘mind-reading’ is easy. Creating tools intuitive and useful enough to convince people to wear a headset may prove more difficult. While Muse and Dreem have circumvented this issue by targeting contexts where users are alone and stationary, promoting adoption in further applications will require some creativity.
One avenue entrepreneurs see to capturing a market early is through piggybacking on Virtual Reality (VR) adoption. EEG electrodes can easily be integrated into existing headset models to add a further element of immersion when hooked up to a VR environment.
VR, along with its relative Augmented Reality, is the next frontier for their Muse headsets. Moffat notes; “We’re at the beginning of a major shift in the application of low-cost brain and biosensors… VR biofeedback and neurofeedback are some of the most engaging and powerful experiences that users can have, and it will completely change how people experience gaming.” Neurable, perhaps the most advanced player in this space, already has an EEG device which can be adapted for use with existing VR headsets with minimal modification. As a proof of concept, they recently released the game Awakening, which can be played entirely without handheld controllers. In the near future, hands-free selection screens and menu navigation may become a staple of VR interfaces.
Open(source) Your Mind
The advancement of budding technologies like consumer EEG relies on nurturing independent developer buy-in. For now, there are too many potential applications for the few industry players to pursue. Only by expanding this base will the platform’s software ecosystem become enticing enough for the average consumer. Understanding this need, Neurosky and Emotiv, also sellers of proprietary BCI hardware, are trying to further lower the barrier to entry for future BCI creatives by promoting software development kits. Many of these founders believe neurotech will soon reach an inflection point, where expanding interest and a diverse application space will drive each other’s growth.
Essential to this future are enthusiast networks like NeurotechX, which has 17 chapters worldwide, including one in San Francisco. The collective serves as a forum for sharing news, hosting workshops, and organizing livestream events in an industry whose current size limits its capacity for outreach. Integral to the growth of this and other groups has been OpenBCI– a low-cost open source BCI platform that can even assembled using a 3D printer. Using this tool, ‘neurohackers’ around the world have innovated with minimal financial investment.
Reminiscent of tech’s nascent days, much of the early adopter BCI community of the last decade has consisted of hobbyists tinkering at home or in maker spaces. Increasingly, however, the learning tools are available for those with basic programming experience. Many platforms allow the extraction of their EEG data without requiring written code. NeurotechX even provides ample free educational material, making it easier than ever for anyone with a computer and initiative the chance to leave prints on the fresh soil of emerging neurotech. For anybody with an interest in this budding technology, there has never been a better time to get involved.
Petal to the Metal
Lack of a relevantly skilled workforce has also limited consumer BCI development. Cultivation of hobbyist communities is therefore important for more than just a larger app catalog; many of these enthusiasts- who come from backgrounds as diverse as engineering, design, or neuroscience- go on to work in the field or start their own companies.
Dan Schmitz is one such convert. He and a group of friends first started exploring neurotech with OpenBCI a few years ago and hacked games like Mario Kart to be playable hands-free. Today he is CEO of Petal, a startup focused on creating a hardware-agnostic API (Application Program Interface) for BCIs. Schmitz started Petal to address what he sees as another major roadblock to more widespread industry adoption: the lack of a unified platform. “Customers just aren’t going to buy different headsets to track their fitness, or their sleep, or to help meditate,” he claims, before emphasizing how “single use cases end up being expensive and troublesome for the user.” The expertise required for processing EEG signals is still extremely niche and in short supply. Petal’s API intends to bridge the gap between the diverse BCI hardware ecosystem and developers in other industry spaces. According to Schmitz, Petal works directly with industry clients to “take live EEG input stream, translate it, and read it back to you in whatever way you need.”
Eventually, Schmitz thinks that as more manufacturers begin to work with teams like Petal, we may see the integration of BCI capabilities with everyday tasks, similar to how the ubiquity of smartphones revolutionized nearly every walk of life. “When you’re always connected, you won’t even think about it. Instead of reaching out to open a door, you’ll just think, or raise an eyebrow, and it will open… It’s like a new superpower.” The emerging ‘internet of things’ will align nicely with this vision. Much of the smartphone app technology designed for use with household items could be repurposed into compatibility with an EEG headset. Petal has already developed prototypes for turning on lights, stoves, and smart mirrors. Right now, tinkering with available APIs, you could even unlock your smart car with your mind. Wholesale integration of BCI functionality with these everyday tasks is lying just around the corner.
Some teams, operating in ‘stealth mode’, are quietly tackling even more complex applications, with aspirations to radically change the current BCI landscape. Neuralink, Elon Musk’s BCI venture, is the most notorious of these stealth companies. It appears to be tackling these complex issues of high-fidelity and high-resolution active BCI, valiantly ignoring the roadblocks inherent to this technology. Whether they will be able to do so noninvasively without a breakthrough in physical science remains to be seen.
Others are striving to make waves through forward-thinking extensions of passive BCI principles. Kernel is attempting to create an external hard drive for your hippocampal memory banks to both aid in regular recall and those suffering from dementia-related memory loss. These efforts may eventually comprise the first steps to establishing to a complete connection between human gray matter and a computer. Openwater wants to open the door to more involved medical applications through the development of extremely high-resolution fMRI, revolutionizing neuro-imaging. The science behind these aspirations is in its earliest stages, but the theoretical feasibility and industry interest is there.
The human brain is one of the last untapped frontiers on remaining on Earth. One of our greatest challenges as we explore its depths will be translate to its inner workings into actionable, meaningful improvements in our lives. With the wealth of creativity and technical knowledge within the Bay Area, it is destined to be an epicenter of the upcoming revolution. As Dan Schmitz says of local enthusiasts everywhere: “Its going to be the most passionate people who want to learn and discover who are going to take us to the next level.” Today, you could hack your Dreem to tell your alarm to sound again if it detects you falling back asleep. Tomorrow, this functionality may come prepackaged.
Nicholas Gentry, B.S. is a 3rd year Biomedical Sciences PhD Candidate in the Ying-Hui Fu lab at UCSF. He studies the relationships between sleep, genetics, and neurodegeneration.