How Diet Affects Memory and the Hippocampus

Picture a man who cannot form new memories. He eats a full lunch. An experimenter leaves the room, returns minutes later, and offers him another lunch. He eats that one too. Offer a third, and he eats a third. He has no idea he has already eaten twice. Yet when asked to rate his hunger on a scale of 0 to 10, he says something like a seven (83:01). His stomach knows he is full. His brain does not know what that fullness means.

This is the kind of experiment USC (University of Southern California) neuroscientist Scott Kanoski uses to make a strange point to Andrew Huberman, host of the Huberman Lab podcast: hunger is not only a signal from the gut, it is a memory problem. The hippocampus, the brain structure most associated with memory, turns out to be doing double duty as a kind of accountant for meals, tracking what came in, when, and how much, so the body can decide whether it is really time to eat again (31:03).

The Body's Memory of Food

Start with GLP-1 (glucagon-like peptide-1), the hormone behind drugs like Ozempic and Wegovy. In its natural form, GLP-1 is released from cells in the intestine around mealtime and acts locally, right where the vagus nerve, the long nerve linking gut to brain, touches the gut wall (14:29). This is a paracrine signal: the hormone does not travel far, it taps the nearest receptor. If that vagus connection is cut, or its GLP-1 receptors are silenced, natural GLP-1 stops reducing food intake, and animals simply eat bigger meals (17:40).

The drugs are a different animal. Ozempic and its relatives raise GLP-1 activity roughly a thousandfold above natural levels (04:57), stay in circulation far longer, and, in at least some formulations, may reach the brain directly rather than working only through the gut-vagus route, though this is genuinely debated among researchers (09:22). That difference matters for side effects, and for durability: Kanoski notes that a majority of patients discontinue the drugs within two years (67:51), mostly from nausea, so whatever long-term brain benefits these drugs might have remain mostly theoretical.

The vagus nerve threads through nearly everything else in the conversation. It does not send one clean signal. Stretch receptors in the stomach report distension, separate chemical sensors in the intestine report nutrients, and these signals move through different fibers at different speeds (75:52). Kanoski's lab has traced where this mixed signal lands: from the brainstem it reaches a structure called the medial septum, where neurons release acetylcholine, a brain chemical tied to learning, directly into the hippocampus (76:34). Using fiber photometry, a method for recording a single brain chemical in real time, his group watched acetylcholine rise while rats ate, dip when they paused, and end a meal higher than where it started (78:40). That rising floor, Kanoski thinks, is the brain tagging the meal as worth remembering.

"We think that these satiation systems are very much tuned into the memory systems in the brain." (Scott Kanoski, 31:26)

The effect shows up in people, too. In a study by psychologist Suzanne Higgs, people ate lunch, and hours later some were prompted to actively recall that meal before a cookie taste test. The group that recalled lunch ate fewer cookies.

"The memory of recent eating will contribute to eating less at the subsequent eating occasion." (Scott Kanoski, 27:56)

A companion study, from researcher Jeff Brunstrom, used a bowl of soup secretly refilled or drained by a hidden pump. Right after eating, people's hunger tracked what they had actually swallowed. Hours later, their hunger ratings flipped: they went by what they believed they had eaten, not what their stomach had actually processed (29:10). Memory outvotes the body once enough time passes.

What Changes the Signal

If meals are memories, then anything that damages the hippocampus should also distort appetite, and diet turns out to be a direct route to that damage. Kanoski's early work used young rats given free access to a sugar solution at roughly the sugar concentration of a soft drink, alongside otherwise healthy food. As adults, those rats showed memory deficits specific to hippocampal-dependent tasks, and the deficits persisted even after the sugar was removed from the diet (35:28). A broader Western diet, about 35 percent of calories from fat and 15 percent from simple sugar, produced similar hippocampal damage in as little as one week of exposure in some protocols (34:40). Other kinds of memory, the sort that do not depend on the hippocampus, were comparatively spared.

"The hippocampus is somewhat of a canary in the coal mine." (Scott Kanoski, 35:02)

A second, less obvious stressor is heat itself. Baking or roasting food triggers the Maillard reaction, the same chemistry that makes a seared steak or a roasted carrot taste better than its raw counterpart. One byproduct of that reaction is a class of compounds called advanced glycation end products, or AGEs, already linked to diabetes and aging. In unpublished work Kanoski describes as a preprint, a study shared before formal peer review, rats fed an AGE-enriched version of an otherwise healthy diet during adolescence showed adult memory deficits (39:56). The effect tracked a specific gut bacterium: a strain called Lactococcus nearly disappeared with heat treatment, and restoring it reversed the memory problem (90:41). A separate strain, Parabacteroides, rose in the sugar-fed rats and tracked with worse memory; artificially boosting it, even without any sugar exposure, produced similar deficits (89:53). Microbiome research is young enough that a PubMed search for it now returns roughly 250,000 results, up from about 7,000 a decade ago (93:05), and the field only had a baseline genetic catalog of gut bacteria after the Human Microbiome Project concluded around 2012 (92:43).

None of this is fixed in place. Kanoski's lab has shown that aerobic exercise can reverse hippocampal memory deficits caused by a prior Western diet in rats, an effect at least partly driven by changes in the brain itself rather than just improved metabolism (48:25). Disrupting the vagus nerve looks like a genuine risk factor on its own: a small study of patients in Taiwan who had surgical vagus nerve cuts for stomach ulcers decades ago showed a higher rate of later dementia (51:36), though Kanoski calls this a correlation worth following rather than a settled finding. Even the social side of eating runs through the hippocampus: oxytocin, usually described as a bonding hormone, reduces food intake when it acts in the brainstem, but Kanoski's lab found the opposite effect in the hippocampus, where oxytocin signaling is required for the well-documented tendency to eat more around familiar people (118:21).

Listening to Kanoski, the overall picture is less a diet plan than a correction to an old mental model. The hypothalamus still runs the basic switches for hunger and fullness. Layered on top is a slower, more interpretive system, routed through the vagus nerve and the hippocampus, that treats every meal as an event worth filing away. Ultra-processed food, which now supplies something like 60 percent of calories in the typical American diet (63:08), and a national obesity rate above 40 percent (101:23), look less like failures of willpower and more like a filing system under strain. Ozempic and its relatives can quiet the noise from that system. They have not, so far, addressed the question of what good information gets filed in its place.