← Back to list
Lex Fridman PodcastPodcast24 May 2021Source: lexfridman.comHost: Lex Fridman

#186 – Bryan Johnson: Kernel Brain-Computer Interfaces

In plain words

Bryan Johnson says his company Kernel is building a wearable, non-invasive brain scanner (like a helmet) to measure brain activity at scale, aiming to make it as common as phones. He's optimistic about the field, predicting both non-invasive and invasive tech (like Neuralink) will grow together in 7 years. Key mentions: Kernel itself (he invested $53M but only 1 of 228 investors backed it); Neuralink (rival, requires surgery); and Braintree (his previous payment company, sold to PayPal, which funded this venture).

AI SummaryAI-generated · may contain errors · verify against the original

Bryan Johnson, founder and CEO of Kernel, discussed in the Lex Fridman podcast the Kernel-Flow brain-computer interface device developed by his company, which can monitor and record brain activity. The core argument is that by measuring brain signals with high bandwidth, the mysteries of the human m

~11 min full read · 7 sections
Deep Analysis

At a Glance

Bryan Johnson (founder and CEO of Kernel, previously founded Braintree and acquired Venmo) articulated his core philosophy on the Lex Fridman podcast: The human mind is the largest yet unquantified raw material in society. Kernel's non-invasive brain-computer interface devices aim to collect brain data at a population scale, thereby formally introducing cognitive engineering into social infrastructure.


Theme 1: Kernel's Technical Path – Non-Invasive, Wearable, High Bandwidth

Johnson believes the core value of brain-computer interfaces lies in measurement rather than control, and Kernel's choice of an optical non-invasive path is key to making brain imaging mainstream.

  • Technical Principle: Kernel Flow uses time-resolved spectroscopy. Each of its 52 modules contains one laser and six sensors, emitting photons at approximately 100 picoseconds. By analyzing the scattering and absorption of photons in the cerebral cortex, it reconstructs blood oxygen level dependent (BOLD) signals, sampling whole-cortex activity across over 1,000 channels.
  • Comparison with fMRI/EEG: fMRI is noisy (120 decibels) and creates a claustrophobic environment; EEG is uncomfortable to wear. Kernel Flow can be worn in natural settings (e.g., during a podcast recording), allowing the user to "record brain activity in the environment of your choice."
  • Divergence from Neuralink's Path: Johnson initially researched invasive technologies as well. However, after a conversation with a neurosurgeon with 30 years of experience, who said, "About every 15 years, there's a new technology that changes everything. It might already be here; you just can't see it yet," Johnson "bet the company" and pivoted to non-invasive approaches. He uses an internet analogy: Kernel Flow is a "blurry but full-screen movie," Kernel Flux (MEG technology) is "1080p full-screen," and Neuralink is "4K but only showing a circle" — each has its trade-offs, and within the next seven years, both technologies will jointly drive ecosystem development.

> "The most significant contribution that Kernel technology has to offer would be the formal introduction of the formal engineering of cognition as it relates to everything else in society."

> This means: The most important contribution of Kernel technology is the formal introduction of the engineering of cognition into every aspect of society.


Theme 2: Pattern Discovery — The "Drake Equation" Strategy and "Zeroth Principle" Thinking

Johnson rejects the "killer app" mindset for planning the future of brain-computer interfaces, instead adopting an ecological discovery strategy akin to the Drake Equation.

  • Funding Dilemma: Johnson self-funded $53 million to launch Kernel, then approached 228 investors, with only one agreeing to invest. The core obstacle was "What is the killer app?" — but Johnson believes this is the wrong question.
  • Drake Equation Strategy: Analogous to the early 1990s internet, the key is "deploying enough devices in the ecosystem so that enough people can play, experiment, and make mistakes, including both experts and non-experts." Kernel's goal is "to compress 20 years of value discovery into 2-3 years."
  • Zeroth Principle: Johnson proposes the "Zeroth Principle" as a complement to first-principles thinking. First-principles thinking holds that "what remains after eliminating the impossible is the truth"; the Zeroth Principle focuses on "insights with civilization-altering properties that do not currently exist in our shared cognition" — such as the discovery of zero, Einstein's concept of spacetime, or AlphaGo's moves. He believes that as computational intelligence becomes ubiquitous, the frequency of zeroth-order insights will increase dramatically, and the form of conscious existence 20-50 years from now may itself be a zeroth-order insight.

> "We cannot predict the future with first principle thinking. That cannot be our imagination set. The future of our conscious existence, 20, 30, 40, 50 years, is probably a zero."

> Meaning: We cannot predict the future solely through first-principles thinking. The form of conscious existence 20-50 years from now is likely a zeroth-order insight.


Theme 3: Self-Quantification — "Firing" the Conscious Mind, Letting Data Decide

Johnson views his methodology in diet and health management — "firing his own conscious mind" — as a preview of future brain-computer interface applications.

  • 200+ biomarkers: Tested every 90 days, covering blood panels, microbiome, food-induced inflammatory responses (cytokine expression), neuroendocrine system, neurotransmitters, micronutrients, etc. Data generates a shopping list: "I never ask my conscious mind what it wants to eat; I only ask what my body reports."
  • The link between sleep and willpower: Internal research shows that deep sleep is independently correlated with impulse control — "If I don't get enough deep sleep, I can't resist temptation the next day." Johnson schedules his last meal at 8:30 AM (fasted after exercise), ensuring no digestive process before sleep, bringing his resting heart rate down to 42 bpm and maximizing HRV.
  • The philosophy of "firing": Johnson believes the conscious mind is "wrong" in most cases — "I assume my conscious mind is wrong in certain key aspects upon landing." He cites his own multiple "reality shifts," from religion to depression to behavioral psychology, stating, "I don't trust reality; I don't trust any given reality."

> "I fired my mind entirely from being responsible for constructing my diet. I never actually ask my mind what it wants. It's entirely what my body is reporting that it wants."

> Meaning: I have completely fired my mind from being responsible for constructing my diet. I never ask my conscious mind what it wants; it is entirely determined by the data my body reports.


Theme 4: Consciousness Engineering and the Future — From Brain Data to Goal Alignment

Johnson views brain-computer interfaces as the infrastructure for "consciousness engineering," with the ultimate goal of achieving "goal alignment" at both the individual and societal levels.

  • The Social Construction of Consciousness: Johnson agrees that "consciousness is largely social" — "I create your consciousness through interaction with you, and I create my own consciousness through interaction as well." Brain data can expand the dimensions of interpersonal communication: from the low bandwidth of text/emojis to the high bandwidth of a "50-dimensional vector space model," thereby generating "deeper empathy."
  • The Closed Loop of GPT-3 and Brain Data: Johnson envisions feeding human brain data back to AI as a training set, "teaching it politeness, friendliness, and conversational propriety" — but also notes the need for "goal alignment" to prevent different versions of AI (socialist, anarchist, polite, party version) from manipulating one another.
  • Privacy and Data Sovereignty: Johnson criticizes the internet's initial rule as "whoever gets the most information wins," and argues that brain data should follow the principles of "individual consent, individual control, and transparent use" — users should be able to delete data, know who is sharing it and under what conditions, and set usage time limits.

> "The real game going on is that evolution built a system of scaffolding intelligence that produced us. We are now building intelligence systems that are scaffolding higher dimensional intelligence. The meaning of life becomes goal alignment."

> In other words: the real game is that evolution built the scaffolding of intelligence that produced us. We are now building intelligence systems that scaffold higher-dimensional intelligence. The meaning of life becomes goal alignment.


Mentioned Positions

Position Guest Stance Key Data
Kernel (Non-invasive Brain-Computer Interface) Founder/CEO, fully committed Self-invested $53 million; only 1 out of 228 investors agreed; team of ~100 people (36-37 PhDs); developed two technologies: Flow (spectroscopy) and Flux (MEG)
Neuralink Peer/Competitor, different approach Invasive technology; Johnson and Elon Musk explored early collaboration but it did not materialize; both sides invested over $100 million
Braintree (Mobile Payments) Founder, already sold Founded in 2007; acquired Venmo in 2012; acquired by PayPal/eBay the same year; clients include Uber, Airbnb, GitHub, Basecamp
Venmo Acquired by Braintree Innovation: made private financial transactions public
Coinbase Early partner Braintree was the first payment company to formally integrate cryptocurrency (2013)
Fitbit Mentioned as analogy Acquired by Google for $2.5 billion; no additional ecosystem value
Apple Watch Mentioned as analogy Device value + ecosystem value + multi-data stream value-add

Judgments Worth Remembering

1. Johnson: The human mind is the largest unquantified raw material in society

We can quantify stars, calories, and genomes, but not our own minds—yet the mind is "the largest raw input material for society at any given moment."

2. Johnson: Discovery in brain-computer interfaces should adopt a "Drake Equation" strategy, rather than searching for a "killer app"

The core is to deploy enough devices, let enough people play and make mistakes, and compress 20 years of value discovery into 2-3 years.

3. Johnson: The Zeroth Principle is a necessary complement to First Principles

First Principles deal with "known knowns," while the Zeroth Principle addresses "insights that currently do not exist in our cognition"—such as the discovery of zero, Einstein's concept of spacetime, and AlphaGo's moves.

4. Johnson: The cost of designing, manufacturing, and distributing intelligence is approaching zero

Evolution once handled the design, manufacturing, and distribution of intelligence; now humans have taken over—when costs approach zero, existence itself becomes a "goal alignment problem."

5. Johnson: Consciousness is, to a large extent, a social construct

"I create your consciousness through interaction with you, and I also create my own consciousness through interaction"—brain data can expand interpersonal communication from low bandwidth (text/emojis) to a high-dimensional vector space.

6. Johnson: Brain data should follow the privacy principles of "personal consent, personal control, and transparent use"

Users should be able to delete data, know the conditions of sharing, and set usage time limits—this is a correction to the internet's initial rule of "whoever gets the most information wins."

7. Johnson: Sleep is one of the most powerful health interventions

Deep sleep is independently correlated with impulse control; the timing of the last meal directly affects resting heart rate and sleep quality.

8. Johnson: The essence of advice is a mirror of the giver; the receiver should understand its "assumption stack" rather than follow it directly

"Listen to advice, but do not adopt it"—the value lies in understanding the knowledge assumption system behind the advice, not the advice itself.