Health Protocols Should Start With Adaptations, Not Routines

Andrew Huberman frames health advice as a search for durable principles rather than a fixed list of behaviors. A protocol matters less, in his account, than understanding what the protocol is trying to accomplish and then choosing a tool that fits a person’s circumstances.

His example is cortisol. Cortisol is commonly reduced to “the stress hormone,” but Huberman describes it as an energy-deploying hormone: it mobilizes glucose and helps produce wakefulness. In his account, the cortisol awakening response is part of why people wake up, alongside falling melatonin and low adenosine. Bright light early in the day, ideally sunlight, increases that morning cortisol rise; he says caffeine, exercise, cold exposure, hydration, and even some stress can push in a similar direction.

The larger objective, as Huberman presents it, is a daily pattern of high cortisol early and low cortisol at night. A higher morning peak, he says, helps set up a lower evening baseline, which supports melatonin and sleep. A stressful event at night is less disruptive if it produces a brief bump on a low baseline than if it lands on an already elevated one.

That turns the usual advice about morning light into a more flexible rule. Morning sunlight may be a highly effective tool, but it is not the rule itself. Someone on a plane, in Scandinavia during winter, or working an unusual schedule may need a different implementation. The target remains an alert, energized beginning to the waking period and a low-arousal transition into sleep.

Huberman applies the same reasoning to exercise. Someone may prefer Pilates to resistance training, or dislike cold showers, but the choice of tool should not obscure the underlying adaptation being sought. The value of public discussion around health, he argues, is that it is beginning to produce principles that can survive differences in geography, taste, schedule, and equipment access.

The distinction also matters for neuroplasticity. Raising brain-derived neurotrophic factor, exercising, taking a drug, or otherwise increasing the biological conditions that support change does not itself create the change. For a circuit to reorganize, Huberman says, there must be precisely timed neural activity: the person has to practice, learn, or confront the thing they want to change.

A chemical state can open a window. It cannot choose the lesson.

Huberman’s account of psychedelics rests on a practical distinction: an intervention may make change more possible, but it does not specify the change or perform the work afterward. Preparation, support, integration, and a concrete direction for learning matter alongside the compound itself.

Andrew Huberman describes early recreational use of LSD and psilocybin as a mistake. He had what he calls a bad trip while young and unsupported, an experience that made him wary of psychedelics for years. His later use was clinically supported and directed toward longstanding psychological trauma.

He first worked with MDMA in a protocol resembling the MAPS model: preparation beforehand, support during the session, and therapy afterward. Across four sessions separated by months or longer, he found it helpful for accessing emotions and understanding what whole-body feelings were about. He calls MDMA an empathogen rather than a classical psychedelic, and says its effects include simultaneous increases in dopamine and serotonin as well as a large increase in oxytocin. But he notes that oxytocin alone does not replicate MDMA’s clinical effects.

For Huberman, MDMA’s useful feature was not euphoria. It made difficult material feel more approachable: “I’ve got myself held,” as he puts it. Yet that same capacity carries a risk. Because MDMA can make nearly anything seem meaningful, peaceful, or lovable, a person can become attached to thoughts that will not retain the same importance after the session. That is one reason he emphasizes having a trained guide present.

Psilocybin, by contrast, exposed the structure of his thinking. Huberman says his own mind naturally uses analogies and symbolic representations, and high-dose psilocybin made recurring patterns visible: forms of self-seduction, for example, or enchantment with future outcomes that were not realistic. Journaling afterward mattered because the useful task was to translate those perceptions into observations and actions available while sober.

The experience was not gentle. Huberman describes psilocybin alone as terrifying and physically difficult, involving purging and vomiting. But he says it revealed problems that ordinary reflection had not made legible. Research by Romain Nardou and colleagues on psychedelics and social-reward learning provided a related frame: psychedelics can reopen a critical period for social-reward learning in the animal work they reported. Huberman’s point was broader and more practical. A heightened state of plasticity has to be paired with the experience, practice, or emotional work that gives the change direction.

The most consequential case he describes combined MDMA followed a few hours later by high-dose psilocybin—what Tim Ferriss calls a “hippie flip.” Huberman says he has done that twice, with clinical support. In his experience, MDMA provided enough self-compassion to enter material that psilocybin made difficult to avoid.

The subject was not simply what had happened to him. It was what had not happened: needs, losses, and experiences he had wanted but could not create for himself at the time. He had done substantial therapy and journaling, and had worked with MDMA and psilocybin separately, but says that category of loss had not emerged as a direct target.

The session was brutal, he says, but it left him with a concrete orientation. What he lacked earlier in life could now become something to build. More than a year later, he says, that work had left him with a durable direction rather than a lingering crash.

Huberman does not turn that experience into a recommendation that people combine drugs. He calls these compounds powerful and precarious, says some people should not take them, and repeatedly emphasizes clinical support. The narrower lesson is that neuroplasticity becomes useful when it is directed toward something specific—including problems a person may not yet be able to identify through ordinary reflection.

Ferriss extends the caution. Psychedelics are not a panacea, he says; he compares the process to complicated reconstructive surgery, followed by demanding rehabilitation. He says that people with bipolar disorder, a family history of schizophrenia, or borderline personality disorder are generally poor candidates under current clinical-trial criteria, and he cautions against cannabis taken with or near psychedelic drugs.

He also rejects the notion that wide psychedelic access naturally produces social peace. In his account, traditional use in many cultures was limited to a small portion of the population, and substances such as ayahuasca had uses beyond treatment, including hunting and warfare.

Both men place psychedelics within a broader category of interventions that may create a learning window. Ferriss discusses an accelerated transcranial magnetic stimulation protocol enhanced with D-cycloserine, an antibiotic used in this setting as a neuroplasticity-enhancing agent. An open-label case series of the ONE-D protocol examined a single-day regimen of optimized, neuroplasticity-enhanced TMS for depression.

Ferriss says the approach compresses a schedule that had already been reduced from several months to five days into one day. He disclosed that he is involved with Ampa Health, co-founded by Jonathan Downar, and reported that the treatment reduced his generalized anxiety and OCD symptoms by roughly 90% for three to four months at a time. He adds that some people have reported longer durability, but does not present those reports as a general expectation.

The shared model is straightforward: medicines, stimulation, exercise, and other interventions can change the conditions under which the nervous system learns. They cannot substitute for the specific learning, rehabilitation, practice, or life changes that make the new state endure.

Andrew Huberman presents his training philosophy as a practical reconciliation between hypertrophy research and the demands of an actual week that includes running, intervals, and recovery. It was sharpened by training with Dorian Yates, the six-time Mr. Olympia associated with a low-volume, very-high-intensity approach also linked to Mike Mentzer and Arthur Jones.

Yates’s method, as Huberman describes it, begins with warm-up and practice sets, then moves to a small number of work sets taken to genuine failure. For a major body part, the structure might be two or three exercises and two hard sets per exercise, with occasional assisted repetitions, drop sets, or rest-pause work after failure. The workout can be short, but the effort is extreme.

Huberman argues that one all-out set, as Mentzer advocated, is not enough for many people. Reaching actual failure requires unusual focus, muscular recruitment, and willingness to sustain discomfort. A second hard set, or added work after failure, makes the method more usable. Training with Yates, he says, clarified how often people mistake stopping under discomfort for true failure.

The tension is with a simple reading of hypertrophy research. A review by Brad Schoenfeld and colleagues challenges the conventional “repetition continuum,” proposing that muscular adaptations can be obtained across a wide range of loading zones. Huberman treats that work as valuable, but objects to extracting a weekly prescription from studies that often operate against a background of little or no cardio.

In his account, the familiar recommendation of roughly 10 to 20 weekly work sets per muscle group and frequent stimulation misses collisions between demands. Someone who performs a hard leg session, runs, and then does high-intensity intervals on an air bike has put substantial stress through the legs even if they directly train them only once. That indirect work may not be a hypertrophy session, but it affects recovery and total load.

The direct resistance sessions generally take him about an hour. At 50, Huberman says, he continues to get stronger on this schedule, though not dramatically from week to week. The system works, in his view, because the training is intense enough to matter and sufficiently spaced to recover from.

The exact equipment is secondary. Huberman favors belt squats because the load rests at the hips rather than on the shoulders, and he uses leg extensions and hamstring curls alongside them. On torso days, he uses pressing, rows, chins, dips, and other compound movements. His warning for anyone copying a Yates-style program is to avoid reducing it to a handful of isolation exercises: leg extensions and curls are not substitutes for hard compound work.

He also includes neck, core, and lower-back training for strength and stability. Huberman uses a four-way neck machine and urges people with no history of neck training not to rush into high-velocity or aggressive work. He recommends slow movement, nasal breathing, and avoiding a jutting chin during lateral flexion; he says spinal-surgeon friends have warned him that poor technique can aggravate the jaw and create cervical-disc risk.

The cardio structure follows the same logic: choose something that permits maximal effort without injury. Huberman prefers short intervals on an assault bike rather than a Norwegian 4x4 because, for him, the upper body fails before the legs during longer hard efforts. The air bike gives his arms, lats, and trunk a secondary stimulus after torso training, but he does not treat it as a hypertrophy workout.

The larger point is not allegiance to a split or machine. It is to account for all stressors, distinguish direct from indirect training, retain compound movements, and make recovery part of the program rather than an afterthought.

Andrew Huberman begins with a definition: peptides are short chains of amino acids; insulin itself is a peptide. The relevant question is not whether a substance is “natural” because the body makes a version of it, but what happens when a synthetic form is delivered at concentrations, routes, and schedules that differ radically from normal physiology.

He credits physician Abood Bakri with a useful framing. Medicine has long searched plants for pharmacologically useful compounds; it is increasingly searching the body for substances that can become medicines. Peptides sit within that trend. But they do not form a coherent class in terms of effect, data quality, or risk.

The first category is GLP-based weight-loss drugs. Drugs that act on GLP can produce very large increases in GLP signaling and substantial weight loss. Earlier GLP drugs were used for diabetes but did not generate the same weight-loss effects as newer agents that drive much larger increases. Huberman notes associated muscle loss, nausea in some users, reduced alcohol craving and perhaps other cravings, and discussion of apathy.

He identifies retatrutide as a particularly powerful compound because it acts across GLP, GIP, and glucagon systems. It may spare some muscle relative to other options, he says, though he still argues that resistance training matters. He also notes emerging lower-dose use, in part to pursue benefits while reducing nausea.

Yet demand has created a gray market. Compounds labeled “for research purposes only” can be legally sold for nonhuman research, but Huberman says many purchasers are plainly injecting them. The concern is not only legal status; it is uncertainty over purity, identity, and contamination. In his account, retatrutide is easy enough to synthesize that people are already using it before formal release.

The second category is growth-hormone secretagogues: compounds intended to stimulate growth hormone release through the brain-pituitary axis. Huberman names sermorelin, ipamorelin, tesamorelin, CJC variants, and MK-677. He says these substances may produce three- to eight-fold increases in growth hormone and IGF-1, but he remains unconvinced that the results justify the complexity for most people.

His own short experiment with sermorelin increased slow-wave sleep but substantially reduced REM sleep and raised his prostate-specific antigen, or PSA. He did not see meaningful fat loss or strength gains, and his PSA returned to normal after stopping. He cannot explain the mechanism and does not plan to repeat the experiment.

Huberman’s personal conclusion is that secretagogues did not seem worthwhile for him. He contrasts them with prescribed pharmaceutical-grade growth hormone, which he says may have more reliable sourcing than poorly characterized peptide products. That is a comparison of sourcing and expected effects, not a claim that growth hormone is broadly appropriate or low risk; he explicitly says he is not telling people to begin taking it.

The third category is tissue repair, especially BPC-157 and TB-500. Huberman has used BPC-157 briefly after strains or pain and says he felt recovery was faster. But he is equally clear about the central limitation: a recovery outcome after an injury is not a controlled experiment, and BPC-157 acts systemically rather than only at the injection site.

The animal findings he cites include cartilage regrowth, fibroblast migration, angiogenesis, and nerve regrowth. No published human trial exists, he says. His own judgment is that BPC-157 is more plausibly a time-limited intervention for a defined injury or perhaps a gut issue than a daily recovery supplement. Ferriss makes the same methodological point from experience: after elbow surgery, he used the BPC-157/TB-500 combination commonly called the “Wolverine stack,” but cannot say whether it helped.

That principle is central to their safety discussion. Huberman says he is not aware of adverse events that can be clearly traced to many non-FDA-approved peptides. Ferriss answers with a car analogy: a long-running model may have many documented accidents because it has enormous real-world exposure, while a new vehicle may have no recorded deaths because almost nobody has driven it. Sparse adverse-event reporting is not a clean bill of health when use is clandestine and follow-up is inconsistent.

Ferriss also stresses injection hygiene. He has had a staph infection in his elbow and warns against reusing needles, reinserting a used needle into a vial, injecting without cleaning the site, or treating self-injection casually. A capsule is not automatically safer than an injection, either; route of administration changes the risk profile rather than eliminating it.

Huberman’s miscellaneous category includes Epitalon and pinealon. He describes Epitalon as a peptide associated with possible circadian and visual-repair effects in animal work. His own experience—a possible improvement in jet lag when combined with morning sunlight—remains uncertain. He also reports that fasting during long-haul travel seemed to help him adjust more easily, perhaps because meal timing affects peripheral clocks in the liver and gut.

Pinealon, despite its name, is not derived from the pineal gland; Huberman says it originates in the cortex. He reports unusually large increases in REM sleep with occasional subcutaneous use—up to three hours in his own case—without apparent loss of slow-wave sleep. He describes that as a personal observation, says he monitors his blood work, and remains interested in occasional rather than routine use.

The more general issue is how people assign trust. Huberman says the growing acceptance of injectable GLP drugs has made injections feel more ordinary, while the word “peptide” can make a synthetic compound seem more intuitively acceptable than a conventional prescription drug. He sees a large consumer market among women as well as men, including interest in low-dose weight-loss drugs, GHK copper, and combinations of peptides.

That cultural shift does not resolve the decision. The questions are more concrete: Is there a defined problem to solve? Is the product reliably what it claims to be? Is there meaningful human evidence for that indication? Can outcomes and adverse effects be monitored? And is the likely benefit large enough to justify the plausible downside?

Huberman is attracted to the prospect of cleaner supply chains, verified products, and more frequent blood work. Ferriss takes a more cautious position: when possible, he prefers something with decades of research and a clearer adverse-event profile over the newest compound with sparse reporting.

Tim Ferriss and Huberman see artificial intelligence as potentially useful less because it will replace doctors than because it can help individuals synthesize data that no single clinician has time to integrate. Ferriss describes friends with elite medical care who uploaded genomes, wearable data, and years of lab PDFs to large language models and surfaced what he calls material, actionable insights that had not previously been noticed.

The claim is not that these systems are complete or infallible. Ferriss’s point is that an individual may be able to devote more attention to their own records than even a conscientious clinician can, and a model can search across disparate files for relationships worth investigating.

Andrew Huberman imagines AI-assisted health management as a system that can make sense of blood work, sleep trends, food intake, supplementation, and disease-specific triggers. He gives the example of someone with multiple sclerosis using such tools to identify dietary or behavioral patterns that worsen symptoms. He is particularly interested in personalized supplementation: recurring blood tests could inform tailored morning and evening packs, with later testing showing whether markers moved in the intended direction.

The more ambitious future, he says, lies beyond models trained on internet text. The internet is an output of human neural activity, not a direct measurement of it. Huberman recounts a conversation with UCSF neurosurgeon Eddie Chang about building tools around recordings of brain and nervous-system activity: identifying the neural signature of a person’s most focused state, observing how sleep or stimulants alter that state, and adjusting behavior or pharmacology accordingly.

Current wearables are a beginning, not a direct nervous-system readout. Devices such as Oura, Whoop, and Eight Sleep mainly infer state from heart rate, motion, and related measurements. Heart-rate variability, Huberman says, reflects in part the activity of vagal pathways that slow the heart. That is useful, but it remains indirect.

He expects sleep to be an early area where systems not only read bodily signals but actively alter conditions around the body. Temperature-regulating sleep systems already adjust the bed environment dynamically. He imagines comparable closed-loop interventions during waking life: changing temperature, posture, sensory input, or stimulation in response to a person’s physiological state.

Sleep interests him because onset, duration, movement, and some physiological correlates can be measured, even if the measures remain imperfect and individual responses vary. A study of slow bed rocking found more stage-N2 sleep and more sleep spindles during two hours of vestibular stimulation, though not a clear whole-night sleep or memory advantage among already good sleepers.

Huberman connects that work to research on carrying crying infants. A Current Biology study described five minutes of carrying followed by five to eight minutes of sitting before laydown as an intervention that soothed crying infants and could promote sleep. For Huberman, the broader question is whether vestibular input helps people disengage from a stable sense of body position as they fall asleep.

He also describes a sleep mask from Nocturnal, a company that says it can measure eye movements and use subtle stimulation behind the ears to induce sleep in under six minutes. Huberman says he has no relationship with the company and is interested because the outcome is concrete enough to test against sleep measures and user experience.

Ferriss’s first response is that such stimulation might keep him awake. Huberman says the sensation can be adjusted until it is barely perceptible and may take acclimation. Neither presents the device as a settled solution.

The larger possibility is a transition from broad, weak wellness devices to interventions that read physiology, alter sensory or neural input, and adjust in response. Huberman is skeptical of many consumer vagus-nerve stimulators because, in his view, they either do not stimulate the vagus nerve or do not do so robustly or specifically enough. But he sees the direction clearly: health technology will increasingly aim not just to describe a state, but to influence it.

Huberman’s book, Protocols, is organized around sleep, exercise, stress, light, neuroplasticity and learning, and personal development. Its aim, he says, is not simply to prescribe behaviors but to explain the science and let readers adapt tools to their own circumstances.

Its final chapter extends that approach to a less measurable problem: how to organize responsibility, ambition, self-discipline, and an honest connection to one’s own desires.

Andrew Huberman had considered calling the chapter “relationship to self,” but its subject is broader. He found a concise formulation in Jungian psychoanalyst James Hollis: “shut up, suit up, and show up.”

Huberman initially heard it as simplistic, even stereotypically boomerish advice. Hollis’s meaning was more demanding. “Shut up” is not a command to suppress distress; it is a call to gratitude and perspective. Someone else is suffering more, and being alive is itself something to recognize when life is difficult. “Suit up” means preparing oneself—not merely tidying a room, but developing the processes and internal condition required to meet one’s obligations. “Show up” means fulfilling one’s roles, which Huberman regards as central to self-worth.

But Hollis’s counsel has a second half. A person also needs regular time outside the cycle of stimulus and response, away from the ceaseless demand to perform and fulfill roles. Ideally daily, or at least several times a week, Huberman says, a person should make room to encounter what Hollis calls the heart’s desires.

That can be uncomfortable. It may reveal that a long marriage, career, or identity is not what someone wanted—or that a person has lived by inertia for so long that they no longer know what they want. Yet avoiding that encounter can produce its own tragedy: a life organized around competent role fulfillment but detached from the self that is supposedly being served.

Huberman does not resolve the tension by choosing introspection over responsibility. The work is to do both: fulfill the roles that sustain a life, and create enough quiet to discover whether that life is genuinely one’s own.