Huberman, Andrew

neuroscientist, Stanford University; host of the Huberman Lab podcast

Andrew Huberman: Sleep, Dreams, Creativity, Fasting, and Neuroplasticity

Lex Fridman Podcast · Feb 28, 2021

• Andrew Huberman
• circadian rhythm and adenosine
• sleep consistency versus duration
• non-sleep deep rest
• testosterone and effort
• breathing techniques for exertion
• intermittent fasting
• REM sleep and dream therapy
• psychedelics and neuroplasticity
• acetylcholine and skill learning
• eye movement and blinking as signal
• science communication

Core ideas

Huberman explains sleepiness as the product of two interacting systems: adenosine, which accumulates the longer one stays awake and binds receptors to create sleepiness independent of time of day, and the circadian clock, a roughly 24-hour temperature oscillation (lowest around 3–4 a.m., peaking in late afternoon) generated by clock genes present in every cell and synchronized body-wide by the master suprachiasmatic nucleus via systemic temperature — a mechanism established by Joe Takahashi's work. He explains "temperature minimum" (roughly two hours before one's natural, alarm-free wake time) as a key circadian landmark: bright light exposure before it delays the clock (later sleep/wake times), while light after it advances the clock (earlier sleep/wake times), with most people's daily light exposure on both ends of the day keeping the cycle stable near 24 hours. He notes humans reliably do better on diurnal than nocturnal schedules by health metrics (immune and metabolic function), tracing this to ancient predator-prey dynamics where nocturnal vulnerability was costly, especially for children.

On sleep quality and quantity, Huberman pushes back against what he sees as excessive cultural anxiety about sleep, arguing that consistency of total sleep duration matters more for performance than total duruation itself (citing a Harvard organic-chemistry-exam study by Bob Stickgold and Emily Hoagland), that sleep and wakefulness are organized into roughly 90-minute ultradian cycles (so waking near the end of a cycle, e.g. six hours rather than seven, can feel better than "more" sleep that ends mid-cycle), and that stress and self-judgment about imperfect sleep can be more harmful than the sleep loss itself, since a study he cites found positive anticipation about the next day predicts good sleep quality even with reduced sleep. He recommends a target room/body temperature drop of 2–3°F to support falling and staying asleep, describes non-sleep deep rest (NSDR) — including hypnosis-based relaxation protocols developed with colleague David Spiegel — as a scientifically supported way to restore cognitive and motor function (citing a Danish study showing 20-minute NSDR resets basal ganglia dopamine to post-sleep levels), and disputes the idea that lost sleep can't be "recovered," distinguishing true sleep debt (falling asleep unintentionally during the day) from mere fatigue from other causes.

Prompted by Lex's upcoming 48-mile run with David Goggins, Huberman discusses effort physiology in depth: epinephrine (adrenaline) drives alertness and effort but has diminishing returns and a "quit point," while dopamine — synthesized from tyrosine and a precursor to epinephrine — sustains willingness to keep exerting effort, meaning genuine enjoyment or gratitude (not "hacks," but real biochemical states) replenishes the very resources needed for sustained effort. He explains testosterone as making effort itself feel good (synthesized from cholesterol via a pathway that competes with cortisol production, so chronic anger/stress can deplete testosterone over time even without sleep loss directly lowering it), contrasts this with prolactin (associated with quiescence and caregiving) and estrogen (associated with emotional equanimity), and describes practical real-time breathing techniques for endurance effort — longer/more forceful exhales to calm a racing heart via the heart-breathing feedback loop (the sinoatrial node's response to diaphragm movement), and "double inhale, long exhale" patterns to reinflate collapsing alveoli and offload CO2 during steady-state exertion — while cautioning there's no secret universal breathing trick beyond these basic real-time adjustments.

On fasting and nutrition, Huberman cites Satchin Panda's peer-reviewed work showing time-restricted feeding (limiting intake to a consistent daily window aligned with daylight) improves metabolic markers and reduces disease in mice regardless of total calories, while noting human evidence is less direct and adherence is a major confound in nutrition science generally. He explains that fasting or low-carbohydrate states raise alertness and epinephrine (an evolved mechanism to motivate food-seeking), while carbohydrate-rich, high-tryptophan foods (rice, turkey) promote sleepiness partly because digestion itself diverts blood to the gut and triggers vagal signaling toward rest; he notes he personally eats carbohydrates in a defined 6–8 hour window rather than doing pure one-meal-a-day eating, which he found left him too food-loaded and sleepy. He emphasizes electrolytes (sodium, potassium, magnesium) as essential for nervous system function during fasting or heavy exertion, cautioning that over-hydration without adequate electrolytes can be dangerous, and discusses (with appropriate caveats about small samples and anecdote) reports from extreme athletes and special-forces contacts describing unusually clear, high-performance states during multi-day fasts, framing this as plausibly linked to a shift toward ketosis when done properly.

The conversation covers dreaming and REM sleep at length: Matt Wilson's rodent (and later primate/human) work shows spatial "place cell" firing sequences are replayed during sleep, encoding emotionally significant locations and events; early-night, slow-wave-sleep dreams tend to be emotionally flat and coincide with growth hormone release, while REM dreams (concentrated toward morning, accompanied by muscle paralysis/atonia and very low epinephrine) carry heavy emotional and "theory of mind" content (a concept from autism researcher Simon Baron-Cohen). Huberman frames REM sleep as a kind of "self-induced therapy" that uncouples emotional intensity from the memory of difficult past events — citing Susumu Tonegawa's hippocampal replay research — and draws an explicit parallel to EMDR and ketamine-assisted trauma therapy, both of which aim to separate traumatic memories from their physiological/emotional charge without erasing the memory itself.

On psychedelics, Huberman discusses Johns Hopkins' rigorously funded, IRB-approved clinical research (referencing Lex's own prior conversation with Matthew Johnson) as a hard-won return to serious academic legitimacy after the stigmatized 1960s era (invoking Ken Kesey and Timothy Leary as cautionary examples of researchers who became their own uncontrolled experiments), and highlights a UC Davis Nature paper on chemically modified psychedelics that preserve neuroplasticity effects while removing the hallucinogenic component. He frames psychedelic-induced neuroplasticity as a "state, not a goal" — opening a window for change without specifying what changes are made — and discusses MDMA's unusual simultaneous elevation of dopamine (pursuit/reward) and serotonin (contentment) as scientifically novel and potentially significant for depression treatment, while stressing that integration work after the plasticity window (not the acute experience itself) is where lasting benefit is actually built, quoting a colleague: "better living through chemistry still requires better living." He connects this to a broader interest in non-verbal, "wordless" states (jiu-jitsu, surfing, dreaming) as scientifically important, renewing, and distinct algorithmic modes of brain function separate from language-based cognition.

On neuroplasticity and future tools, Huberman highlights Michael Merzenich's (and colleagues Greg Recanzone and Michael Kilgard's) research showing that acetylcholine release from the nucleus basalis, paired precisely with a sensory or motor event, permanently reorganizes cortical maps — a mechanism he believes is underused for accelerated skill, language, and motor learning despite strong evidentiary support, and speculates safe pharmacological (e.g., nicotinic) augmentation paired with structured practice could eventually be combined with brain stimulation (TMS, focused ultrasound) to deliberately direct plasticity. He argues neuroscience currently lacks unifying "conceptual leadership" compared to fields like genomics, but is optimistic that computer vision, machine learning, and robotics will help decode fundamental neural algorithms (e.g., neurons that multiplex functions, as seen in retinal cells switching roles between day and night vision) faster than biology alone could. He and Lex also discuss eye movement, blinking, and pupil dynamics as an underexplored but promising non-contact signal of cognitive/autonomic state (noting blink rate drops with alertness and rises with fatigue, and that blinking itself resets subjective time perception), and reflect on live, unrecorded platforms like Clubhouse as creating a distinctive intimacy in voice-only human connection, closing with mutual reflections on the demands of high-output, ambitious work (using Elon Musk as a case study) and the tradeoffs of starting new, more public or entrepreneurial ventures.

Excerpts