This chapter covers how the brain actually decides what you eat — the gustatory pathway that makes "tastes good" a central judgment rather than a mouth sensation, the three converging signals (gut sensing, metabolic accessibility, belief) behind craving, and the conditioning mechanics that let preference be rewired. It then turns to the compounds proposed to support neuron structure and cognition, including creatine, whose brain effects the chapter's own sources put at doses several times the standard muscle protocol. It closes with sauna as a heat-dose intervention with an inverted-U risk curve and a proposed heat-shock-protein mechanism for its dementia-risk association.
Start with the wiring, because it explains everything that follows. Receptors in the tongue and mouth detect five taste categories — bitter, sweet, umami, salty, sour. Chemical binding converts to an electrical signal, which the gustatory nerve carries to the brainstem's nucleus of the solitary tract and from there to the insular cortex. That destination is the important part: the insula is the same region that perceives internal body states — gut pressure, acidity, the general interoceptive read on what is happening inside you.
Because taste terminates in an interoceptive region rather than staying peripheral, "this tastes good" is a centrally generated judgment about internal state, not a surface sensation in the mouth. The mouth supplies data; the verdict is issued elsewhere, alongside signals that never passed through the tongue at all. This is the structural reason food preference behaves less like a fixed property of a food and more like a learned conclusion about it (see Taste Perception, Gustatory Pathway & Preference Conditioning).
Not all of it is learned. Some taste responses are largely hardwired — children's innate preference for sweet, the aversion to putrid food. But most others are soft-wired and malleable through experience, and that malleability is the entire basis for deliberately changing what you like. The hardwired residue matters too: strong innate "yuck" responses stay largely resistant to the retraining methods described later in this chapter, so preference change has a ceiling.
Huberman's framework holds that food choice and craving are driven by three converging signals rather than by taste alone, and that evaluating a food decision means checking all three axes rather than just flavor or calorie count (Three-Signal Model of Food Choice).
The first is subconscious gut-neuron sensing. Neuropod cells lining the gut are specialized neurons precisely tuned to amino acids, sugars, and fatty acids. They report food quality to the brain via the nodose ganglion, triggering dopamine release that drives food-seeking — and they do this independently of conscious taste. You can be pulled toward a food by a signal you never tasted.
The second is metabolic accessibility. On this account, the brain's underlying seeking mechanism targets how metabolically active or accessible a food makes neurons — not taste, not dopamine, and not blood glucose directly. "Craving" is then reframed as an indirect readout of that deeper drive rather than as the drive itself. This is also why sweet foods are liked: not simply for the taste, but because sweetness predicts a metabolic response. Food reward and food reinforcement are distinct concepts, and separating them is what makes preference recalibratable at all.
The third is belief — what a person believes a food contains or will do. The milkshake study is the demonstration: belief about a food's calorie and nutrient content altered physiological measures, including insulin and blood glucose, independently of the food's actual content. The bodies are explicit that this is a belief effect and not a placebo effect; placebo shifts symptom perception, whereas this shifts a measurable physiological signal.
If liking is a learned central judgment, it can be conditioned — in both directions, deliberately or by accident.
The accidental direction first. Repeatedly pairing an artificial sweetener with a glucose-raising food in the same eating occasion conditions the sweetener alone to later trigger both dopamine release and an insulin/blood-glucose response. This is straightforwardly Pavlovian: the sweet taste becomes a predictor, and the body starts responding to the prediction. The recommended fix is equally simple — consume artificial sweeteners apart from glucose-raising foods within an eating occasion, so the pairing never forms (Taste Perception, Gustatory Pathway & Preference Conditioning).
The second accidental effect is preference narrowing. Progressive consumption of very sweet or highly palatable foods narrows the dopamine system so that it rewards mainly those foods. The consoling half of that claim is that the same system can be recalibrated toward less-sweet and less-savory foods, which it will then reward instead — the narrowing is a setting, not a permanent loss.
The deliberate direction uses the seeking mechanism from the Three-Signal Model of Food Choice as leverage: eat a less-palatable healthy food alongside a stimulus that shifts brain metabolism, and the learned positive association forms around the healthy food. Roughly 7–14 days of sustained exposure to a new food pattern is described as enough to shift subjective taste preference. Two limits are worth naming plainly. The bodies never specify what concretely counts as "a stimulus that shifts brain metabolism," so the technique is stated at a higher level than it can be executed at; and strong innate "yuck" responses remain largely resistant regardless of exposure.
Huberman highlights six compounds as supporting neuron structure and cognitive function, each with dietary sources and, where applicable, a dosing threshold (Brain-Support Nutrient Compounds & Dosing). These are function-specific micronutrients and precursors, deliberately distinct from the macronutrient targets covered in Protein & Muscle-Building Nutrition.
All six are obtainable from food. Supplementation is framed as optional — a way to reach higher levels rather than a requirement — the same food-versus-supplement optionality framing applied to other supplement classes in Longevity Interventions. Huberman discloses that he personally relies on EPA, creatine, and alpha-GPC supplementation as a "baseline insurance policy" despite not actively eating much fish or seeking dietary creatine.
Creatine is where this chapter's material argues with itself, and the disagreement is worth reading carefully rather than smoothing over.
One position, the one embedded in the six-compound list, is that peer-reviewed research shows creatine improves neuronal and brain function and fuels frontal cortical circuits tied to mood and motivation — and that a minimum of 5 g/day is a sensible baseline for anyone, regardless of meat intake (Brain-Support Nutrient Compounds & Dosing).
The other position is that 5 g/day is a muscle dose that says little about the brain. On this reading, creatine's brain and cognitive benefits are concentrated in metabolically or cognitively stressed populations — sleep deprivation, hypoxia, jetlag, aging 60+ — and require substantially higher doses than the standard protocol, closer to 20 g/day for at least a week in the best available studies. The proposed bottleneck is the blood-brain barrier: astrocytes guarding the barrier appear to have limited or no creatine transporter expression, unlike the blood-side endothelial cells, so the low blood creatine levels produced by a 5 g dose may not reach the brain effectively.
A second, competing mechanistic story appears alongside it: creatine does cross the blood-brain barrier, but at 5 g/day muscle tissue preferentially consumes it first, so brain-relevant effects may require doses well above 5 g — 10–20 g/day in cited protocols. Note that these two explanations are not the same claim. One says the barrier blocks it; the other says muscle outcompetes the brain for it. The chapter carries both without reconciling them, though they converge on the same practical advice.
What both share is the framing that makes creatine a useful case study: dosing thresholds matter more than compound choice, and the proposed benefit is stress-conditional — surfacing under sleep deprivation, psychological stress, aging, or neurodegenerative disease rather than at baseline. That is a different shape of claim from most brain-support compounds discussed for general use. For the full mechanism, the safety evidence, and the evidence-quality caveats behind these numbers, see Creatine Fundamentals; for the yardstick these claims should be held to, see Evaluating Evidence & Supplement Quality.
Regular sauna use — 4–7x/week — is associated with a 66% lower dementia and Alzheimer's risk than once-weekly use, and with reduced cardiovascular disease risk. The evidence was strong enough to convert a stated skeptic: Peter Attia moved from skepticism to advocacy after reviewing roughly five years of data (Drive Podcast ep. 369, Attia/Galpin) (Sauna Use for Cardiovascular & Cognitive Health).
The proposed mechanism runs through heat. Sauna's cardiovascular benefit partly mimics moderate aerobic exercise, and it is dose-dependent on heat shock protein activation — roughly a 50% increase over baseline at 163°F for 30 minutes. HSPs prevent protein misfolding and aggregation and remain active well after the session ends; animal studies show they protect against amyloid-beta aggregation, which is the proposed bridge to the dementia-risk reduction.
More heat is not linearly better. There is an inverted-U risk relationship with temperature: benefits plateau around 180°F, while extreme temperatures — 200°F and up, to 212°F, of the kind seen on social media — are associated with increased dementia risk rather than added benefit. Both podcast speakers lowered their personal sauna temperatures on the strength of this. It is the same dose-response discipline applied elsewhere in this corpus to protein and creatine, and the same reason the creatine dosing argument above is a threshold question rather than a more-is-better one.
Format matters for reaching the dose at all. Infrared saunas raise heart rate and core temperature more slowly than dry saunas, so roughly double the session length — about 40 minutes versus 20 — is needed to approximate the same cardiovascular stimulus as a standard dry-sauna protocol of 163–190°F for 20–30 minutes.
One further pathway is flagged as suggestive rather than conclusive: a single hyperthermia treatment raising core body temperature by about 2°F (Charles Raison's work) produced a six-month antidepressant effect versus sham control in prior research. That points to a mental-health effect of heat exposure distinct from its cardiovascular and cognitive effects, but the chapter's own framing stops well short of treating it as established.
The through-line is that the brain's evaluations are computed, not given. Taste is a central judgment assembled in an interoceptive region (Taste Perception, Gustatory Pathway & Preference Conditioning); craving is an indirect readout of a metabolic-access drive, and it can be moved by gut sensing and by belief alone (Three-Signal Model of Food Choice). That the machinery is computed is exactly what makes it retrainable on a 7–14 day timescale.
The second through-line is dose-response with a wrong answer on both sides. Sauna has an explicit inverted-U — too cool does too little, 200°F+ is associated with worse dementia outcomes (Sauna Use for Cardiovascular & Cognitive Health). Creatine's brain story is a threshold argument, where the standard 5 g may simply fail to reach the target tissue (Brain-Support Nutrient Compounds & Dosing). And both creatine's brain benefit and sauna's HSP protection are framed as showing up under stress — sleep deprivation, aging, protein-aggregation load — rather than as baseline improvements in a healthy, rested person.
Where the material is genuinely thin, it is worth saying so rather than filling the gap. Phosphatidylserine is listed with sources and a supplement form but no dose at all. The anthocyanin mechanism is stated as unresolved — direct neuronal effect or inflammation reduction, unknown — with only a soft "a cup or two fairly often" as guidance. Glutamine's 1–10 g/day is a tenfold range attached to a "consult a physician first," which is not a protocol. And nothing here addresses how these compounds interact with each other, or whether heat exposure and the nutrient list combine, compete, or are simply unrelated levers on the same outcome. The bodies never claim to have tested that combination, and this chapter should not either.