UCSF study reveals how digestion helps the brain learn to love coffee, chocolate, fat, and sugar

A flavor can become a favorite long after its first appearance on the tongue. New research identifies a brain pathway that helps explain that change, linking the taste of food with the nutritional consequences that emerge as digestion begins.

Scientists at UC San Francisco found that nutrients reaching the gut trigger a sustained dopamine signal in a small part of the amygdala. In mice, that signal helped teach the brain which flavors predicted nourishment. Human brain scans provided evidence of related activity in the same region.

The study, published in Nature, was led by James C. R. Grove and senior author Zachary Knight, a UCSF physiology professor. Collaborators included Dana Small of Yale University and Marc Tittgemeyer of the Max Planck Institute for Metabolism Research. “We've shown that this slow dopamine system is what drives us to learn to like certain foods over time, and we can see where and how that happens,” Knight said.

Coffee offers a familiar illustration of an acquired taste, although the experiments did not test coffee or caffeine. Instead, they examined how the brain learns about calories. The findings extend research on gut–brain communication and the distinction between flavor and food reward.

Dopamine helps the brain learn what matters and which experiences are worth pursuing. Many familiar dopamine responses unfold within seconds, such as those triggered by a rewarding taste or an external cue. Nutritional feedback presents a different challenge because its consequences arrive more slowly.

To separate those signals, Grove and colleagues delivered glucose or a liquid nutritional formula directly into mouse stomachs. This bypassed the mouth, allowing the researchers to measure responses to ingested nutrients without the immediate contribution of taste. They compared dopamine release across several brain regions.

A strong response appeared in the anterior basolateral amygdala, or aBLA. With the nutritional formula, dopamine release began about 48 seconds after nutrients reached the stomach. It remained elevated for roughly nine minutes after the infusion ended, revealing a signal that outlasted the initial delivery.

The result builds on earlier work identifying dopamine subsystems that track internal states. Separate research on gut-driven sugar preference likewise showed why tasting something sweet and detecting its nutritional content are different biological events. The new study traces how those events support learning.

The researchers next asked whether dopamine was necessary for learning which flavors predicted calories. Mice received two differently flavored drinks. Licking one triggered glucose delivery into the stomach, while licking the other triggered water delivery. Later, the animals chose between the flavors without those nutrient infusions.

After six days of training, control mice strongly preferred the glucose-associated flavor. Its share of their licks rose from about 35% before training to 81% afterward. When researchers silenced dopamine-producing neurons in the ventral tegmental area during training, the mice failed to acquire that preference.

The intervention did not prevent consumption of a familiar nutritious food. It also did not stop mice from learning to avoid a flavor associated with illness. Silencing dopamine-responsive cells within the aBLA similarly blocked nutrient-related preference learning, supporting a specific role in updating the value of flavors.

Those distinctions matter because learning to like nourishment differs from learning to avoid poisoning. A separate study of delayed illness feedback examines the latter process. The findings also add a neural mechanism alongside broader investigations of how the gut influences food choices.

The team then reversed its approach. Mice tasted calorie-free flavors, and one flavor was paired with light-based stimulation of dopamine fibers in the aBLA. The stimulation followed licking by 20 seconds, approximating the delay built into the stomach-infusion experiments. No nutritional reward accompanied the stimulated flavor.

Nevertheless, the mice learned to prefer it. Stimulating comparable dopamine projections in several other tested brain regions did not produce the same effect. This showed that activating the aBLA pathway could substitute for nutrient feedback in that particular flavor-learning task, rather than merely accompanying it.

The response was also distinct from a general drive to seek stimulation. Mice did not press a lever to activate these fibers, and stimulation did not increase overall food intake. The slow dopamine pathway therefore appeared specialized for teaching a flavor’s nutritional value under the tested conditions.

The researchers also imaged individual aBLA cells and found responses to both taste and later nutrient signals. This helps explain where the association might form. Related work distinguishes the sensory and nutritional effects of sugar, while the new study identifies a circuit that joins those kinds of information.

Nutrients produced a larger sustained dopamine response in the left BLA than in the right in mice. Simultaneous recordings from both sides confirmed the difference. This asymmetry raised another question: Would a similar pattern appear in humans, whose food preferences involve much more than nutritional feedback?

The team reanalyzed PET data from 10 healthy men who received a milkshake or a noncaloric control drink. The data supported dopamine release in the BLA, but the authors said the sample was too small for a properly powered comparison between hemispheres. A separate experiment addressed that question.

In that fMRI experiment, 38 adults consumed glucose and underwent measurements of blood-oxygenation changes associated with neural activity. The post-ingestion response was stronger in the left BLA. Unlike PET, fMRI did not directly measure dopamine release, and the experiment did not establish that this activity caused human flavor learning.

The human evidence supports a shared anatomical pattern, while the strongest causal evidence comes from mice. Grove and colleagues’ study also points toward further investigation of connections with the insular cortex. The broader food-reward literature shows why several interacting systems must remain part of the explanation.

Mapping the slow dopamine pathway gives researchers a more precise target for studying how preferences develop. The team also connected the response to intestinal nutrient sensors. Blocking a glucose transporter prevented the glucose-triggered dopamine response; blocking several fat sensors eliminated the response to fat infusion.

That does not yet provide a method for changing anyone’s diet. The study did not test GLP-1 medications, and Knight’s suggestion that these drugs might alter nutrient-related reinforcement remains a hypothesis. Establishing such an effect would require experiments specifically measuring the pathway during treatment and assessing resulting changes in learning.

The research also leaves coffee’s particular appeal unresolved. Caffeine, aroma, repeated exposure and social experience are different from the glucose-driven learning studied here. Earlier investigations of sugar sensing and internal-state dopamine signals provide context without proving that one pathway explains every acquired taste.

What the study offers is a clearer account of one fundamental process: learning from the consequences of eating. A meal provides information twice, first through its flavor and later through its effects inside the body. This pathway helps the mouse brain connect those experiences and adjust what it seeks next.

These resources explore how internal signals shape food learning, dopamine activity and reward. Entries appear from newest to oldest and exclude the current study.

A neural mechanism for learning from delayed postingestive feedback
This mouse study examines how delayed illness signals reactivate flavor representations in the amygdala and support food-avoidance learning. (Nature, April 2025)

Dopamine subsystems that track internal states
This earlier study identifies dopamine populations that track different stages of food and fluid ingestion, providing a foundation for the new work. (Nature, July 2022)

The gut–brain axis mediates sugar preference
Researchers describe a gut-to-brain pathway that supports sugar preference independently of sweetness sensed in the mouth. (Nature, April 2020)

Rethinking Food Reward
This review examines evidence that nutrient sensing within the body contributes to reinforcement beyond a food’s consciously perceived flavor. (Annual Review of Psychology, January 2020)

A Neural Circuit for Gut-Induced Reward
This study investigates how gut-innervating vagal sensory neurons connect digestive signals with brain reward systems. (Cell, September 2018)

Health disclaimer: This article is for informational purposes only and does not provide medical advice. Consult a qualified health professional about dietary concerns or medication decisions.

Research findings are available online in the journal Nature.

The original story "UCSF study reveals how digestion helps the brain learn to love coffee, chocolate, fat, and sugar" is published in The Brighter Side of News.

Like these kind of feel good stories? Get The Brighter Side of News' newsletter.