When I was approached to join Kernel, a startup building brain interfaces (BI), I was hesitant. I knew that the path to commercialize this technology was (and is) fantastically challenging. After nearly a year of working with Kernel as an advisor, I concluded that joining Bryan Johnson and his remarkable team was a once in a lifetime opportunity, not only to make breakthroughs many consider impossible, but to bring a technology to market that is so fundamentally consequential.

While neuroscience is an integral part of building brain interface (BI) technology, it has become evident that the major breakthroughs needed to make this technology broadly accessible are in solid-state quantum devices, materials science, and photonics. I have been leading research in these areas for the past 25 years and building novel devices and systems in industry, at university, and at two startups. Almost all major research that I have been leading involved multidisciplinary innovations, many of them relating to — and often inspired by — cell biology, human physiology, and human health.

Despite having a fantastic academic research environment at Northwestern University, I decided to join Kernel because it presented a unique opportunity. Kernel is not a conventional start up, where a certain idea or technology is at the core of its activities. Instead, it is a collection of the best talent from different fields working in tandem to achieve the technological and scientific breakthroughs needed for making compact BI with unprecedented performance a reality.

Humanity has been discovering the rules of Nature and using them to make new tools to explore the unknowns of the universe from time immemorial. Yet the brain represents one of the least discovered of all known unknowns. This is not accidental, but simply due to the fundamental challenges involved.

The anatomy of the brain at the macroscopic and microscopic scales is well known, but it turns out that the machinery of the brain heavily relies on nanoscopic parts, which have yet to be mapped out. There is significant progress being made in this endeavor, such as the use of high throughput electron microscopes and the bright x-ray sources at the Advanced Photon Source at Argonne National Lab. However, these developments can only give a static map of the brain, and little information about its dynamics. This is akin to having a map of a city with detail at the level of a grain of salt, but completely frozen in time.

One might think that we could simply map the whole brain in real time to capture its dynamics. However, it turns out that the challenge is somewhat fundamental in nature: to map the brain with great detail, the uncertainty principle of quantum mechanics suggests that we need to use high momentum particles, but these would easily destroy the living cells. Even if we could map the brain in motion, we would face the daunting task of making sense of it all.

Instead of grappling with the whole brain, BI presents a reasonable path for incrementally developing our understanding of the brain. There are a growing number of methods to communicate with neurons in the brain, and the number of connected channels (single units) has increased from a few to thousands over the past few decades.

However, none of the existing published methods are suitable for a compact BI that could be used on the non-patient population, and outside of highly restrictive and controlled environments. In short, a compact non-invasive BI with performance similar to invasive methods has remained elusive despite its great scientific and commercial prospects.

This again is due to some fundamental challenges, including the very limited energy that can be directed to the brain safely, the extremely weak signal levels within the brain due to the near-perfect (quantum and thermodynamic limit) performance of neurons, and the spatio-temporal resolution needed to have a meaningful BI system.

At Kernel, we are realistic about the magnitude of the challenges involved in making a commercially viable brain interface system. It entails making devices near the theoretical limit of sensing, power consumption, weight, size, and cost. It requires efficient and extensive collaboration between experts from widely different disciplines, including physical scientists, neurobiologists, information theorists, hardware engineers, and software engineers.

Most importantly, we know that existing technologies are not sufficient, and that we need to achieve new breakthroughs in science and technology to reach this goal.

Many in the field have illuminated potential applications of BI for future technologies (e.g. mind-controlled prosthetics or typing by brain). However, the potential is much larger than most have imagined.

With history as a guide, two trends will likely emerge: enhanced human-human interaction and enhanced human-machine interaction. The former will dramatically accelerate the expansion of collective human knowledge, while the latter will significantly shrink the time-scale of tasks that it is applied to.

In enhancing human-human interaction, BI will lead to a similar blossoming in human interactions as the invention of the printing press did in the past. Just as the press did, BI will connect many minds across time and space at an exponentially higher scale than before. In parallel, by enhancing human-machine interaction, BI will radically reduce the time for any task. It will produce massive increases in efficiency, similar to the invention of textile mills kicking off the first industrial revolution.

Despite the significant challenges, we are extremely motivated, since producing the first commercially viable BI system is arguably the only path forward. Only such a breakthrough can attract the financial backing, manpower, and infrastructure needed for rapid progress in this field, which will catalyze the next technological revolution.

Hooman Mohseni
Chief Technology Officer at Kernel

(cross posted from Bryan Johnson to t.net with Hooman as the orginal author, posted with permission)