Food, memory and cravings: Kanoski

USC neurobiologist Scott Kanoski joined Huberman Lab on 12 October 2026 to discuss a two-way relationship: food can affect memory, and memory can influence how much people later eat. The episode also covered GLP-1 drugs and sweeteners alongside the vagus nerve and gut microbiome^[1]. The useful part is the framework. The caveat is that several examples came from rats, not clinical trials in people.

Memory participates in appetite

Kanoski's central argument is that eating does more than deliver nutrients and trigger gut hormones. A meal also creates a memory, which means later hunger can depend partly on what a person remembers eating rather than only on calories still being processed. This gives the hippocampus, a brain region central to episodic memory, a role in appetite regulation. Memory matters.

A controlled human experiment cited in the show tested this directly. One hundred volunteers saw either 300 ml or 500 ml of soup before lunch, but a hidden pump changed how much they actually consumed. Immediately after lunch, hunger tracked the amount eaten. Two and three hours later, hunger tracked the amount participants believed they had eaten. Those who remembered the larger portion reported less hunger^[2]. Memory did not erase physiology, but it changed the later subjective signal.

For the first time, this manipulation exposes the independent and important contribution of memory processes to satiety.

GLP-1 appetite effects use several routes

The episode presents GLP-1 signaling as a brain-and-body system rather than a single switch. A preclinical paper from Kanoski and colleagues tested liraglutide and exendin-4 in rats. Blocking GLP-1 receptors in the central nervous system reduced part of the food-intake effect. Cutting vagal sensory input below the diaphragm changed dose responsiveness but did not eliminate suppression at higher doses. The authors concluded that vagal afferents and direct central GLP-1 receptor activation both contributed^[3].

That animal work helps explain why medications such as semaglutide can affect appetite and what patients often call food noise. It cannot determine exactly how much each pathway contributes in a given person. Nor does it show that changing the dose outside a licensed schedule will improve cognition or craving control. The approved labels and human outcome trials remain the right sources for clinical use.

Kanoski also distinguishes the GLP-1 made after eating from long-acting receptor agonist medicines. Endogenous GLP-1 is released from the gut and participates in short-lived meal signaling. A pharmaceutical agonist is designed to activate the same receptor for much longer. Shared receptor biology does not make the natural hormone and a weekly medicine equivalent in exposure or effect. This helps explain why a meal may produce a brief satiety signal while prescribed treatment can alter appetite across days.

The sweetener evidence needs a species label

Kanoski also discussed low-calorie sweeteners. One cited experiment gave adolescent rats one of three sweeteners and followed them into adulthood. The animals later showed impaired hippocampal-dependent memory and altered glucose handling, with changes in sugar-motivated behavior but no weight gain^[4]. The researchers did not find strong microbiome changes, and their brain gene-expression findings varied by region and sex.

This is a mechanistic animal study. It supports a question about early-life exposure and later behavior, but it does not establish that a particular sweetener causes memory impairment in children or adults. The three sweeteners were studied within one rodent model, and real diets differ in dose, duration, food context, and baseline health. A headline that drops the word rats would overstate the evidence.

Food choice and memory form a feedback loop

The conversation links highly processed diets with poorer hippocampal performance, then asks whether weaker meal memory could make later intake easier to trigger. This is plausible as a feedback loop, and animal studies give researchers ways to test its neural wiring. Human nutrition is harder to isolate because sleep, stress, exercise, income, education, and existing metabolic health can travel with diet patterns.

The human soup experiment is valuable because it manipulates meal memory while holding actual intake apart. The rat studies are valuable for testing nerves, receptors, brain regions, and gene expression. They answer different questions. Neither design supports a simple claim that one food rewires memory or that one appetite drug repairs cognition.

Cravings also resist a single-pathway explanation. Learned cues, reward history, current energy state, and the social setting can all affect whether a food feels compelling. Medication may reduce hunger without deleting every learned association. Likewise, better meal recall may influence later intake without overriding sleep loss, stress, or access to food. The episode is strongest when these systems are treated as interacting influences rather than competing explanations.

What the episode does not establish

The discussion does not show that GLP-1 medicines are cognitive enhancers. It does not compare tirzepatide with semaglutide on memory outcomes. It does not prove that artificial sweeteners cause dementia, or that microbiome changes explain every link between diet and behavior. Kanoski's cited work ranges from controlled human feeding experiments to receptor studies in rodents. Keeping those evidence levels separate is necessary for an honest digest.

For readers following semaglutide evidence and regulation, the strongest takeaway is modest. Appetite reflects gut signals and brain processing, while remembered meals can shape later hunger. The episode supplies a research map, not a diet prescription. Questions about medication choice, nutrition during weight loss, or persistent cognitive symptoms belong with a clinician who can assess the whole case.