December 14, 2021 | Edoardo D'Anna, Data Scientist
Why I Joined Kernel...
A screenshot of Kernel Flow’s operator user interface, which is designed to provide essential information during system setup. In this example, the real-time "scalp coupling index” is shown for each of the 52 modules (i.e. each hexagon). The scalp coupling index is an indicator of how well the device is making contact with the head and is a useful guide for fine-tuning headgear placement.
Why I joined Kernel
All views expressed in this article are my own.
I've got some news. In June of this year, I moved to sunny Los Angeles to join Kernel as a Data Scientist. I left academia (again) to join the small group of engineers, scientists, designers and many more who have banded together to change the world by democratizing neuroscience. In this post, I want to share why I joined Kernel and what I've learned in the six months since doing so. I hope to shed some light on what Kernel is attempting to achieve, and perhaps help others who are currently at a crossroads in their professional lives (especially early-career scientists and engineers). So let's get to it!
Why I joined
A compelling mission (hint: it's not about control)
Working in the field of neurotechnology for the last ten years has been an incredibly exciting and rewarding experience, offering a first-row seat to a major shift from academic research and science-fiction towards exciting companies and products, big and small. Perhaps even more notable is the way the field has captured the public's imagination, spurring a myriad of discussions and debates around a wide range of topics, from augmentation to brain uploading, all the way to the nature of consciousness itself.
This earnest effort to turn decades of neuroscientific research into compelling products has spawned a vast (and ever-increasing) number of startups. A sizable share of these privately funded efforts focus on control (e.g. Facebook, Neuralink, Paradromics). The idea is compelling—replace the computer mouse with a mere thought (or flick of the wrist)—but is also limiting. Control is a fairly narrow problem. And the bar is set very high (for non-medical applications). Superseding the intricate machinery that is the hand is a tall order. There certainly will be, and already are, some specific (often niche) use-cases where the benefits of a hands-free neural control signal can make up for most of the limitations, but I think truly mainstream adoption is still somewhere over the horizon. That's not to say I don't believe in a future where control will be entirely mediated by neural interfaces. I think it's inevitable. Rather, the current focus on control may be distracting us from more immediate, and perhaps more exciting, applications.
This is where Kernel comes in. Kernel is openly and intentionally not pursuing control. In fact, the very nature of Kernel's first device, Flow (a functional near-infrared spectroscopy helmet that uses light to measure hemodynamic changes in the brain), makes it a poor choice for replacing your mouse. Instead, Kernel is going after something else: quantifying the brain.
It's been famously said that "what gets measured gets improved." Without dwelling on the merits of that particular quote, it does outline an important fact: we are blind to what we cannot measure. If your doctor doesn't know your blood pressure, they can never hope to fix it. Similarly, counting your steps or measuring your resting heart rate can lead to significant improvements in fitness and health. Regrettably, when it comes to the brain, we measure very little. Sure, we have expensive and difficult to interpret medical imaging. But we lack simple, widespread and easy to understand measures of brain activity and health. The brain doesn't have the equivalent of a blood pressure cuff (i.e. something you might measure once or twice a year in the right setting, like your doctor's office ). And the brain most certainly doesn't have the equivalent of a fitness tracker (i.e. something you might use at home, every day).
That's fundamentally Kernel's ambition: measure something of substance about the brain non-invasively and distill it into a metric that's as simple as possible, yet still informative. Even a single brain measure fitting the above description would be a game-changer—enabling us to inform, entertain, improve, guide, learn and grow. That's a clear and exciting mission, and one I was immediately drawn to.
Openness and humility
As a scientist and lifelong skeptic, I have been fairly transparent about my dislike of the neurobable and hyperbole that is prevalent in the neurotech industry. While I can certainly appreciate the necessity for embellishment, I think there's a fine line between sales tactics and plain old nonsense—one too many happily step over.
Since I first started following Kernel, I was struck by the openness and humility with which they approach their ambitious goals. I have watched as Kernel leaned into and cultivated this philosophy, and I now consider it to be one of its greatest strengths. Internally, this approach ensures that decisions are made rationally and that scientific rigor is embraced. Externally, it builds trust within the neurotech community and beyond and serves as a template for other companies to follow. This thinking is also reflected in Kernel's commitment to giving users full agency over what happens to their personal and brain data—a refreshing position in today's tech landscape.
A recent and compelling example of this philosophy at work is the Flow U program, designed to put Kernel's device in the hands of scientists for a free year-long trial. If researchers don't find value in what Kernel has to offer, they can return the device at no cost.
Importantly, while Kernel's approach is marked by openness and humility, it does not lack in determination when it sets its sights on a target. Nor does it shy away from taking credit for its successes—which have been remarkable, as we'll touch upon later. To me, this represents the perfect balance and allows Kernel to set groundbreaking, yet realistic, ambitions.
A word about the academia-industry dichotomy
I had resolved not to stoke the flames of this infamous debate any higher. But the thought that it might help others—including my younger self—set a course in their professional lives with more confidence made me reconsider.
Some say the differences between industry and academia are overplayed, yet my experience has been that of two consistently different worlds. A lot of the obvious distinctions have become well-worn clichés by now and need no repeating (i.e. the money, work-life balance, etc). There is one, however, which I have encountered less often, yet was the most meaningful for me personally: the level and depth of collaboration.
Companies offer a much more collaborative, open and sharing working environment. The best explanation for this I can think of is that academia is ultimately a single-player game, where you—the researcher—are the brand and the product. My experience outside of academia has been starkly different. The goal alignment intrinsic to laboring together towards a clearly articulated goal lends itself to much more spontaneous, frictionless and genuine collaboration—and on much larger scales. While this is certainly a matter of personal preference, the feeling of belonging and cooperation stemming from this type of working environment can be highly fulfilling, and frankly: a lot of fun.
Luck, opportunity, and the windiness of life's path
In addressing the question of "Why I joined Kernel," it would be disingenuous not to acknowledge the role played by luck and opportunity. Yet I think there is an important lesson in this too. Life is hard to predict and plan for. Things go awry, surprises happen (I'm looking at you COVID-19), and reality gently brushes your plans aside.
But amidst the noise of circumstance, it pays to be prepared and it certainly pays to be persistent. In hindsight, I can see that the path that led me to Kernel was a long spiral, rather than a straight line. When I first heard of Kernel's existence, several years ago, I immediately applied. While that disappointingly didn't work out, learning about Kernel's progress—and writing about it on this blog—ultimately put me on a collision course with them. Without necessarily knowing it, the choices I made along the way prepared me to seize the opportunity when it came. Like Steve Jobs famously said: "You can’t connect the dots looking forward; you can only connect them looking backwards. So you have to trust that the dots will somehow connect in your future."
What I learned
A really compelling mission
I've explained why Kernel's ambitious mission was so compelling to me from the outside looking in. What I've learned since joining the company has deepened my appreciation for it. The brain is an unimaginably complex object that we are only beginning to understand. Despite our relative ignorance, we can already measure a remarkable number of things about it using various physical sensing modalities (electrical, optical, magnetic, acoustic, etc). Amongst these non-invasive approaches, Kernel's Flow device stands out by using an advanced optical measuring technique (time-domain functional near-infrared spectroscopy, which I explored in more depth here) that gives it unique advantages over more commonly used methods (such as electrical-signal based EEG).
These precise measurements will allow Kernel to define simple yet informative metrics that quantify the brain's health and state. While a simple low-dimensional measure will never describe the full complexity of a system like the brain, there are clear advantages to breaking down large, difficult-to-probe systems into digestible insights. Some familiar examples of (relatively) complex systems being reduced to accessible metrics include: resting heart rate (an indicator of cardiovascular health), heart rate variability (an indicator of both cardiovascular health and autonomic nervous system function), and body-fat percentage (an indicator of fitness and adiposity). Each of these examples allows entire companies (or divisions) to exist and can be measured at home with affordable and simple to use devices (smart scales, fitness trackers, etc). What if we had a handful of metrics like that for the brain (our most important and perplexing organ)? For instance, what if we had a metric that captured brain performance/ability, and one that captured brain health/aging. Wouldn't you want to know these two numbers for your brain and how they compare to others in your age group, or finally be able to measure the impact of your life choices on its health and performance? I know I certainly would!
These rigorous, yet simple to understand metrics could form the basis of a wide range of compelling applications in areas as diverse as self-improvement (e.g. neurofeedback), entertainment (e.g. adaptive video-games), mental health (e.g. monitoring, virtual reality therapy, etc), learning (e.g. real-time state monitoring), and many more.
While the appeal for individuals is clear, the current price and form factor put that slightly out of reach (although as I've written in the past, Kernel's ambition is to rapidly democratize access to these insights by having a Flow device in every home by 2033). Luckily, the very same metrics I described above can provide tremendous value to individuals without ever putting a device on their heads. We live in the era of social media, targeted ads and behavioral nudging, all of which have profound—yet often poorly understood and measured—impacts on our brains. With the tools at Kernel's disposal, we have the opportunity to identify and communicate how a given activity or product impacts your brain, using simple, validated metrics rather than obscure, difficult to replicate ad-hoc studies. Think of it as identifying fast food for the brain so you can steer clear.
Taking this line of thought further, we can imagine how forward-thinking companies might agree to go through a certification process akin to an organic label for the brain, designed to verify that their experiences and products have a positive or neutral impact on the brain. Armed with this knowledge, consumers might begin making more deliberate choices about what they expose their brains to and companies might finally have the incentive to take mental health and wellbeing into account early in the design process.
All of this is easier said than done. But Kernel's strength isn't that it has all the answers (although we have some ideas 🤫). Rather, Kernel's strength is that it is currently the best-positioned player—by a long shot—to get to those answers. I'll expand on why I believe that to be the case next.
World-leading team and expertise
I've mentioned my affinity for the collaborative spirit of startup companies. At Kernel, this is magnified by the exceptionally talented individuals it is made up of. While my sample size is modest, I can safely say that I have never felt surrounded by as much competence and brilliance as I am at Kernel. One of the most difficult challenges a startup faces is assembling a world-class team while safeguarding its values and unique character. Kernel has done a tremendous job at this.
What is perhaps even more remarkable is that Kernel has achieved this as a full-stack company—one which deals in everything from circuit design, mechanical prototyping, and software engineering to machine learning, neuroscience, cloud analysis, and more. Very few places (companies or otherwise) can say they have a state-of-the-art device that is entirely designed, assembled, tested and then deployed as part of large scale, rigorous neuroscientific research studies—all in the same building.
The quality and breadth of expertise on Kernel's team, perhaps more than any other point I made in this post, is what gives me immense confidence in its ability to succeed.
A device so good that Kernel has no direct competition
Let's talk product. Up until now, I've focused on the intangible: the vision, the mission, the talent. But Kernel has been around for some time, and those qualities have already delivered some very tangible results—Flow.
By all conceivable metrics, Kernel Flow is leaps ahead of any other non-invasive optical system out there (i.e. number of channels, head coverage, sampling frequency, ease-of-use, scalability, etc). More importantly, Kernel has carved out a unique niche within the entire spectrum of non-invasive brain measuring techniques. In contrast to other common approaches, like EEG, Flow has greater spatial resolution and access to unique information, such as the absolute concentration of oxygenated hemoglobin in the cortex. The ability to tap into this rich information—for the entire head—is one of Kernel's key competitive advantages. I do not believe that EEG, especially low channel-count systems (e.g. frontal only, in-ear, etc) will be able to provide the paradigm shift that Kernel is angling for. The only non-invasive devices that can compete with the data Flow produces are multi-million-dollar fMRI machines, which have none of the ease-of-use, portability and scalability needed to truly democratize access to these insights.
On the other end of the spectrum, invasive techniques (which require some form of surgery to insert into the brain) can offer rich information and are uniquely poised to improve the lives of people with spinal cord injury and other forms of trauma and disease of the nervous system. However, when it comes to consumer applications, invasive interfaces' current inability to capture data for the entire cortex, and their obvious barriers to entry, put them on an altogether different—and longer-term—trajectory.
It's therefore clear that Kernel operates in a very unique space. Flow strikes a careful balance: it can capture rich information about the brain while remaining portable, affordable and scalable. Nobody else can quite say that today—Kernel has invented and engineered itself out of any direct competition.
As I've said in a previous post, being on the most promising path still doesn't guarantee success—our mission may turn out to be impossible altogether. What I hope to have imparted, however, is that Kernel is currently the one company in the world best positioned to harvest the fruits of decades of neuroscience and share them widely.
The long road ahead won't be easy
Lest I be labelled overly optimistic—perhaps even naive—let me finish with this. Kernel is a startup. And a startup is still a startup: an unlikely bet taken in pursuit of a better future one believes to have glimpsed. The wilder the future, the riskier the bet. And Kernel is after a pretty wild future.
Everyone at Kernel is well aware that the road that lies ahead is littered with unknowns. There will most certainly be stumbles and a few dead-ends, perhaps even difficult decisions to make—that's the cost of dreaming big.
Kernel's mission of democratizing brain measurement hinges on finding the intersection between rigorous neuroscience, clarity and mainstream appeal. That's a tall order. But I think nobody else has a better chance—and that's an exciting position to be in.