How Food Rewrites Memory, Craving, and Satiety: Scott Kanoski on the Gut-Brain System

1. Guest Background

This Huberman Lab episode is a long interview with an identifiable guest: Dr. Scott Kanoski. Hosted by Andrew Huberman and titled “How Food Choices Shape Memory, Cognition & Cravings | Dr. Scott Kanoski,” it was uploaded on 2026-10-12 and runs 7,326 seconds, or a little over two hours. The subject is not simply nutrition in the everyday advice-column sense. The episode asks how food choices interact with memory, hunger, satiety, GLP-1 signaling, craving, cognition, and the body-to-brain systems that make eating more than calorie intake.

Kanoski is introduced as a professor of human and evolutionary biology and neurobiology at the University of Southern California. His lab studies how hunger and memory interact; how mind and body influence appetite, body weight regulation, craving, and food choice; and how different nutrients affect brain development and later brain function. That background matters because the interview is built around mechanisms. Kanoski does not primarily argue for one named diet. He explains how gut signals, metabolic hormones, vagal pathways, microbiome changes, acetylcholine, dopamine, and hippocampal memory can shape what an animal or person eats next.

The conversation has two distinct voices. Huberman often translates the science into public-health and listener-facing questions: GLP-1 drugs, ultra-processed food, artificial sweeteners, dementia risk, muscle loss, food addiction, and the practical problem of knowing what to eat. Kanoski answers as a mechanistic researcher. Much of his evidence comes from rodent models because those models allow tight control of diet, timing, and intake. He also repeatedly marks uncertainty when the evidence is preclinical, unpublished, correlational, or not yet tested in human randomized trials. That restraint is one of the episode's main editorial strengths.

2. What the Episode Covers

The episode begins with GLP-1 because GLP-1 drugs now sit at the center of public debate about appetite and weight. Kanoski first separates natural GLP-1 physiology from drug intervention. In its natural form, GLP-1 is an incretin hormone secreted from the distal intestines around food ingestion, with some release before and during a meal. There is also a brain source: neurons in the brainstem nucleus tractus produce GLP-1 and send signals broadly through the brain. Peripheral and central GLP-1 both suppress appetite, but Kanoski says they appear to operate independently. GLP-1 also helps regulate blood glucose by stimulating insulin secretion when glucose is high, which is why GLP-1-based drugs were first diabetes treatments before they became famous for weight loss.

From there, the interview moves into Kanoski's core research question: how diet affects hippocampal-dependent memory. He describes a high-sugar model in which juvenile or adolescent animals had free access to a sugar solution comparable in caloric percentage to a soft drink, while still having healthy food and water. Those animals later showed hippocampal-dependent memory deficits that persisted after sugar was removed. He also describes a junk-food model using chocolate peanut butter cups, ruffled potato chips, and high-fat high-sugar rodent diet; memory impairments persisted after animals were switched back to healthy diet, though he carefully notes that this does not mean no other intervention could reverse them. A Western-style diet with roughly 35% of calories from fat and 15% from sugar also impaired hippocampal-dependent memory in animal models.

The most explanatory part of the episode is the gut-to-hippocampus pathway. Kanoski argues that eating and food choice are not simple reflexes mediated only by brainstem and hypothalamus. The hippocampus receives gut-related vagal input and circulating metabolic signals such as insulin, ghrelin, and leptin. His team has identified a three-tier pathway: vagal information reaches the brainstem, then the medial septum, then acetylcholine-producing medial septum neurons innervate the hippocampus. Using fiber photometry, the lab observed hippocampal acetylcholine rising while animals held food, chewed, and swallowed, falling during pauses, and ending the meal at a higher baseline than before the meal. Kanoski thinks this may help animals remember what they ate and where. He adds that healthy food and high-fat high-sugar diet produced similar hippocampal acetylcholine eating profiles, while non-nutritive stevia did not; his working guess is that taste plus post-oral nutrient sensing may be required. The same discussion turns to interoception: Kanoski says internal body signals are unlike visual signals because they combine neural and endocrine inputs into a messy gestalt. He illustrates that with amnesic patients who can eat repeated meals while hunger and fullness ratings still change, and with hippocampal-lesioned rats that fail to interpret internal energy-state cues. High-fat high-sugar diet can also impair awareness of hunger and fullness, which makes interoception a plastic and diet-sensitive process rather than a fixed readout.

The interview then widens to artificial sweeteners, food processing, microbiome mechanisms, dementia risk, dopamine reinforcement, and oxytocin. Kanoski describes rodent work in which juvenile/adolescent exposure to stevia, ace-K, or saccharin within FDA acceptable daily intake levels was followed by adult memory and metabolic changes. He discusses unpublished preprint work isolating food processing by heat-treating otherwise healthy chow to induce Maillard reactions and advanced glycation end products. He explains the VTA-to-nucleus-accumbens dopamine projection as a reinforcement pathway activated not only by palatable food but also by bland food during hunger, water during thirst, and sodium during sodium depletion. Near the end, oxytocin becomes a reminder that feeding signals are context-dependent: brainstem oxytocin can reduce meal size and increase satiation, while hippocampal oxytocin signaling appears necessary for social facilitation of eating in rodents.

3. Core Views: Reasoning, Examples, and Limits

The first load-bearing idea is that food does not merely fuel the brain; it helps determine what the brain remembers. Kanoski's emphasis on the hippocampus is not a generic claim that diet affects “memory” in the abstract. The hippocampus is relevant because eating is a memory-rich behavior. Animals need to remember where food was found, what happened after it was consumed, whether the location is worth revisiting, and whether they have eaten recently. The human examples make the point concrete. In Suzanne Higgs's pizza-lunch work, people primed to remember a recent lunch ate fewer cookies afterward. In Jeff Brunstrom's soup-bowl experiment, hunger immediately after eating was driven by actual intake, but hunger hours later was driven more by how much soup participants believed they had consumed. Satiety, in this account, is not only stomach volume. It is a body signal interpreted through memory.

The second idea is that Kanoski's diet-and-memory case rests on converging animal models, not a single dramatic experiment. The high-sugar, junk-food, and Western-diet models all point toward hippocampal-dependent memory as especially vulnerable to diet quality. The tasks also matter. His lab uses spatial food-location tasks and object-location tasks, not vague claims that animals became globally less intelligent. In some diet manipulations, animals perform near chance while controls perform above chance. That gives the evidence structure. Still, the limitation is built in: rodent work has unusually strong control over what animals eat and how much they eat, but developmental windows, experimental diets, and species differences limit direct translation to human life.

The third idea is that GLP-1 drugs are powerful tools, but the episode refuses to flatten them into miracle drugs. Kanoski accepts that clinical trials support efficacy and that these drugs push GLP-1 signaling into a superphysiological range. He also notes common nausea and vomiting, discontinuation, and the open question of long-term efficacy beyond initial weight loss. Microdosing is especially uncertain: he is not aware of animal or human data and would want large randomized controlled trials, but doubts those will happen soon because studies are expensive and companies have little incentive to test people buying less drug. Even the site of action remains unresolved. Newer drugs such as Ozempic and Wegovy may act in circumventricular organs, may enter the brain more slowly or less clearly, or may have effects that current fluorescent-labeling methods miss. “Food noise” is therefore a report in search of a mechanism, not a solved biological entity.

The fourth idea is that artificial sweeteners and ultra-processed foods should be decomposed mechanistically rather than judged by slogans. Kanoski does not endorse a simple “harmful” or “inert” conclusion for low-calorie sweeteners. In his lab's juvenile/adolescent rodent work, daily exposure within FDA acceptable intake levels was followed by adult memory impairments, glucose regulation changes, larger insulin spikes after glucose infusion, and altered tongue sweet taste receptor expression. But he explicitly marks the rodent-model caveat and says human randomized trials do not provide a definitive good-or-bad answer. His approach to processing is similarly granular. Instead of treating “ultra-processed” as one black box, his team heat-treated otherwise healthy food to induce Maillard chemistry and AGEs. The unpublished preprint status matters; it is a promising mechanism, not settled clinical guidance.

The fifth idea is that the microbiome is a plausible mechanism, not a magic explanation. In the sugar model, Kanoski found a changed bacterial pattern, with parabacteroides elevated about twofold and negatively correlated with memory. Enriching that population after antibiotics caused memory impairments even without sugar exposure. In the heat-treated AGE-enriched model, lactococcus was eliminated, positively correlated with memory, and could rescue impairments when reintroduced. Those are stronger than casual microbiome associations because they attempt to reproduce or rescue behavior. Yet Kanoski is careful: microbiome science cannot yet list clean “good” and “bad” bacteria. A population may look beneficial in one ecological context and harmful or irrelevant in another. The field is young and expanding quickly, which means the current safe conclusion is complexity, not bacterial shopping lists.

The sixth idea is that dopamine reinforcement is state-dependent. The VTA-to-nucleus-accumbens pathway is activated by palatable foods such as sucrose and high-fat sucrose combinations, but also by bland food when an animal is food-restricted, water when thirsty, and sodium when sodium-depleted. Dopamine here is better understood as a reinforcement, motivation, and need-state currency than a pleasure button. This helps explain why food preference can change. Kanoski agrees that after reducing sugar intake, formerly liked sweet foods may become too sweet. The conditioned taste-aversion example shows even stronger plasticity: a sucrose-related dopamine response can disappear when sucrose is paired with malaise. But he also says careful studies have not yet answered whether healthy foods can reach the same reinforcement thresholds as more potent stimuli.

The seventh idea is that public health cannot separate weight, nutrition, and cognition. Huberman worries that GLP-1 drugs may solve the weight problem without solving nutrient quality. Kanoski replies that weight-loss programs should include nutrition and physical activity, and that clinical trials usually embed those guidelines. Dementia is framed with similar caution. Prospective human evidence links midlife dietary patterns, metabolic status, and visceral adiposity with later dementia risk, but those are correlations. In transgenic Alzheimer's animal models, Western high-fat simple-sugar diets can worsen pathology and bring symptoms earlier. The strongest supported inference is not that a specific diet cures dementia; it is that diet quality, physical activity, and metabolic health belong in lifelong risk management rather than last-minute repair after symptoms appear.

4. Learning and Application

The most useful application is to think of eating as a layered brain-body system rather than a simple menu problem. The long-term exposure layer is the most straightforward. In Kanoski's animal models, excessive sugar, high-fat high-sugar diet, and junk-food diet repeatedly impair hippocampal-dependent memory. The practical implication is not to demonize every sweet taste, every fat source, or every processed food. It is to reduce long-term, high-frequency exposure to combinations that repeatedly load the system with simple sugar, high fat, and highly rewarding snack structure, especially during childhood and adolescence when Kanoski's models suggest developmental exposure can have long tails.

The second application is to take meal memory seriously. The pizza and soup studies imply that later hunger is partly shaped by what a person remembers or believes they ate. This supports simple behaviors: notice the meal, avoid turning every meal into distracted background activity, and make the amount and composition of food visually and cognitively available enough that the next hunger signal has context. The boundary is narrow. The episode does not prove that mindful eating replaces calorie management, treats eating disorders, or overrides metabolic disease. It supports the more modest idea that recent eating memory can influence later hunger and likely later intake.

For GLP-1 users, the practical frame is integration rather than substitution. If someone uses a GLP-1 drug under medical supervision, the episode supports treating it as a powerful appetite and metabolic tool, not a complete health plan. Kanoski says weight-loss programs should embed nutrition and physical activity; Huberman worries about nutrient quality and lean mass when appetite falls. Kanoski is less worried about long-term micronutrient deficiency because many users stop within two years and he is not aware of such problems in trials, but he also notes those trials are often one to two years and usually involve diabetic or obese participants. The boundary is clear: reduced appetite is not the same as adequate protein, micronutrients, resistance training, or clinical monitoring. Microdosing should not be treated as evidence-based simply because anecdotes exist.

Artificial sweeteners require a similar middle path. If a low-calorie sweetener replaces habitual sugar-sweetened beverages, some weight-related outcomes may be favorable. But Kanoski's animal work raises a real developmental question: repeatedly separating sweet taste from caloric consequences may alter memory, glucose handling, and sweet-taste processing. For adults, the useful stance is to ask what the sweetener is replacing, how often it is used, and whether it sustains a constant need for intense sweetness. For children and adolescents, caution is more reasonable because Kanoski's developmental models are exactly where the signal appeared. The boundary is equally important: the episode does not establish that these sweeteners inevitably harm human cognition, and it does not establish that they are biologically inert.

Food processing and the microbiome should be handled as mechanisms, not identity labels. Kanoski's heat-treatment work suggests that Maillard reactions, AGEs, and bacterial changes may matter, but the work is unpublished and preclinical. His sugar and AGE models also show that changing one bacterial population can reproduce or rescue memory effects in animals, yet he warns that the field cannot name universal good and bad bacteria. In daily life, that argues against chasing a single probiotic as a brain-enhancement shortcut. More defensible targets are overall diet quality, fiber, fermented foods when tolerated, sleep, stress management, and careful antibiotic use under medical guidance. A fecal transplant example for C. diff does not generalize into a broad cognitive therapy.

Exercise is the clearest companion behavior in the episode. Kanoski says rodent evidence shows exercise can reverse memory deficits caused by Western diet, likely through both metabolic and brain changes. That supports pairing diet improvement with physical activity rather than treating them as separate projects. It also fits his dementia-risk framing: healthy dietary patterns should be lifelong, not an emergency intervention after symptoms appear. The boundary is that neither exercise nor diet is presented as a guaranteed prevention or treatment for Alzheimer's disease. The stronger, supported claim is that they are part of prospective risk management alongside other brain-health behaviors.

Finally, the reward-system discussion gives a realistic way to think about preference change. Kanoski's dopamine account means cravings are not fixed traits. They are shaped by internal state, repeated exposure, learning history, and food environment. Reducing the frequency of very sweet or high-fat high-sugar stimuli can make less extreme foods more rewarding over time; formerly liked sweets may begin to taste too sweet. The tradeoff is that this recalibration is not instantaneous and may not be equally easy for everyone. Kanoski does not equate food addiction with drug addiction, but he does acknowledge that DSM criteria can support food-addiction-like arguments. Some people will need more than individual willpower: structured environments, medical care, family changes, and policy changes may matter.

Source

• Original episode: How Food Choices Shape Memory, Cognition & Cravings | Dr. Scott Kanoski