Scientists have discovered a specific group of brain cells that create memories of meals, encoding not just what food was eaten but when it was eaten. The findings, published today in Nature Communications, could explain why people with memory problems often overeat and why forgetting about a recent meal can trigger excessive hunger and lead to disordered eating.

During eating, neurons in the ventral hippocampus region of the brain become active and form what the team of researchers call “meal engrams” — specialized memory traces that store information about the experience of food consumption. While scientists have long studied engrams for their role in storing memories and other experiences in the brain, the new study identified engrams dedicated to meal experiences.

“An engram is the physical trace that a memory leaves behind in the brain,” said Scott Kanoski, professor of biological sciences at the USC Dornsife College of Letters, Arts and Sciences and corresponding author of the study. “Meal engrams function like sophisticated biological databases that store multiple types of information such as where you were eating, as well as the time that you ate.”

Distracted eating implications

The discovery has immediate relevance for understanding human eating disorders. Patients with memory impairments, such as those with dementia or brain injuries that affect memory formation, may often consume multiple meals in quick succession because they cannot remember eating.

Furthermore, distracted eating — such as mindlessly snacking while watching television or scrolling on a phone — may impair meal memories and contribute to overconsumption.

Based on the experiment’s findings, meal engrams are formed during brief pauses between bites when the brain of laboratory rats naturally survey the eating environment. These moments of awareness allow specialized hippocampal neurons to integrate multiple streams of information.

Kanoski said it can be assumed a human’s brain would undergo a similar phenomenon. When someone’s attention is focused elsewhere — on phone or television screens — these critical encoding moments are compromised. “The brain fails to properly catalog the meal experience,” said Lea Decarie-Spain, postdoctoral scholar at USC Dornsife and the study’s first author, “leading to weak or incomplete meal engrams.”

Mechanism of ‘meal memories’

The research team used advanced neuroscience techniques to observe the brain activity of laboratory rats as they ate, providing the first real-time view of how meal memories form.

The meal memory neurons are distinct from brain cells involved in other types of memory formation. When researchers selectively destroyed these neurons, lab rats showed impaired memory for food locations but retained normal spatial memory for non-food-related tasks, indicating a specialized system dedicated to meal-related information processing. The study revealed that meal memory neurons communicate with the lateral hypothalamus, a brain region long known to control hunger and eating behavior. When this hippocampus-hypothalamus connection was blocked, the lab rats overate and could not remember where meals were consumed.

Eating management implications

Kanoski said the findings could eventually inform new clinical approaches for treating obesity and weight management. Current weight management strategies often focus on restricting food intake or increasing exercise, but the new research suggests that enhancing meal memory formation could be equally important.

“We’re finally beginning to understand that remembering what and when you ate is just as crucial for healthy eating as the food choices themselves,” Kanoski said.

In addition to Kanoski, other study authors include Lea Decarie-Spain, Cindy Gu, Logan Tierno Lauer, Alicia E. Kao, Iris Deng, Molly E. Klug, Alice I. Waldow, Ashyah Hewage Galbokke, Olivia Moody, Kristen N. Donohue, Keshav S. Subramanian, Serena X. Gao, Alexander G. Bashaw and Jessica J. Rea of USC; and Samar N. Chehimi, Richard C. Crist, Benjamin C. Reiner and Matthew R. Hayes from the University of Pennsylvania’s Perelman School of Medicine; and Mingxin Yang and Guillaume de Lartigue from the Monell Chemical Senses Center; and Kevin P. Myers from the Department of Psychology at Bucknell University.

The study was supported by a Quebec Research Funds Postdoctoral Fellowship (315201), an Alzheimer’s Association Research Fellowship (AARFD-22-972811), a National Science Foundation Graduate Research Fellowship (DK105155), and a National Institute of Diabetes and Digestive and Kidney Diseases grant (K104897).

The study found that oleic acid, a monounsaturated fat associated with obesity, causes the body to make more fat cells. By boosting a signaling protein called AKT2 and reducing the activity of a regulating protein called LXR, high levels of oleic acid resulted in faster growth of the precursor cells that form new fat cells.

“We know that the types of fat that people eat have changed during the obesity epidemic. We wanted to know whether simply overeating a diet rich in fat causes obesity, or whether the composition of these fatty acids that make up the oils in the diet is important. Do specific fat molecules trigger responses in the cells?” said Michael Rudolph, Ph.D., assistant professor of biochemistry and physiology at the University of Oklahoma College of Medicine and member of OU Health Harold Hamm Diabetes Center.

Rudolph and his team, including Matthew Rodeheffer, Ph.D., of Yale University School of Medicine and other collaborators at Yale and New York University School of Medicine, fed mice a variety of specialized diets enriched in specific individual fatty acids, including those found in coconut oil, peanut oil, milk, lard and soybean oil. Oleic acid was the only one that caused the precursor cells that give rise to fat cells to proliferate more than other fatty acids.

“You can think of the fat cells as an army,” Rudolph said. “When you give oleic acid, it initially increases the number of ‘fat cell soldiers’ in the army, which creates a larger capacity to store excess dietary nutrients. Over time, if the excess nutrients overtake the number of fat cells, obesity can occur, which can then lead to cardiovascular disease or diabetes if not controlled.”

Unfortunately, it’s not quite so easy to isolate different fatty acids in a human diet. People generally consume a complex mixture if they have cream in their coffee, a salad for lunch and meat and pasta for dinner. However, Rudolph said, there are increasing levels of oleic acid in the food supply, particularly when access to food variety is limited and fast food is an affordable option.

“I think the take-home message is moderation and to consume fats from a variety of different sources,” he said. “Relatively balanced levels of oleic acid seem to be beneficial, but higher and prolonged levels may be detrimental. If someone is at risk for heart disease, high levels of oleic acid may not be a good idea.”

“No genuine sauropod gut contents had ever been found anywhere before, despite sauropods being known from fossils found on every continent and despite the group being known to span at least 130 million years of time,” says lead author Stephen Poropat of Curtin University. “This finding confirms several hypotheses about the sauropod diet that had been made based on studies of their anatomy and comparisons with modern-day animals.”

Knowledge of the diet of dinosaurs is critical for understanding their biology and the role they played in ancient ecosystems, say the researchers. However, very few dinosaur fossils have been found with cololites, or preserved gut contents. Sauropod cololites have remained particularly elusive, even though these dinosaurs may have been the most ecologically impactful terrestrial herbivores worldwide throughout much of the Jurassic and Cretaceous periods, given their gigantic sizes. Due to this lack of direct evidence when it comes to diet, the specifics of sauropod herbivory — including the plant taxa they ate — have been largely inferred based on anatomical features such as tooth wear, jaw morphology, and neck length.

In the summer of 2017, the staff and volunteers at the Australian Age of Dinosaurs Museum of Natural History were excavating a relatively complete subadult skeleton of the sauropod Diamantinasaurus matildae from the mid-Cretaceous period, which was found in the Winton Formation of Queensland, Australia. During this process, they noticed an unusual, fractured rock layer that appeared to contain the sauropod’s cololite, which consisted of many well-preserved plant fossils.

Analysis of the plant specimens within the cololite showed that sauropods likely only engaged in minimal oral processing of their food, relying instead on fermentation and their gut microbiota for digestion. The cololite consisted of a variety of plants, including foliage from conifers (cone-bearing seed plants), seed-fern fruiting bodies (plant structures that hold seeds), and leaves from angiosperms (flowering plants), indicating that Diamantinasaurus was an indiscriminate, bulk feeder.

“The plants within show evidence of having been severed, possibly bitten, but have not been chewed, supporting the hypothesis of bulk feeding in sauropods,” says Poropat.

The researchers also found chemical biomarkers of both angiosperms and gymnosperms — a group of woody, seed-producing plants that include conifers. “This implies that at least some sauropods were not selective feeders, instead eating whatever plants they could reach and safely process,” Poropat says. “These findings largely corroborate past ideas regarding the enormous influence that sauropods must have had on ecosystems worldwide during the Mesozoic Era.”

Although it was not unexpected that the gut contents provided support for sauropod herbivory and bulk feeding, Poropat was surprised to find angiosperms in the dinosaur’s gut. “Angiosperms became approximately as diverse as conifers in Australia around 100 to 95 million years ago, when this sauropod was alive,” he says. “This suggests that sauropods had successfully adapted to eat flowering plants within 40 million years of the first evidence of the presence of these plants in the fossil record.”

Based on these findings, the team suggests that Diamantinasaurus likely fed on both low- and high-growing plants, at least before adulthood. As hatchlings, sauropods could only access plants found close to the ground, but as they grew, so did their viable dietary options. In addition, the prevalence of small shoots, bracts, and seed pods in the cololite implies that subadult Diamantinasaurus targeted new growth portions of conifers and seed ferns, which are easier to digest.

According to the authors, the strategy of indiscriminate bulk feeding seems to have served sauropods well for 130 million years and might have enabled their success and longevity as a clade. Despite the importance of this discovery, Poropat pointed out a few caveats.

“The primary limitation of this study is that the sauropod gut contents we describe constitute a single data point,” he explains. “These gut contents only tell us about the last meal or several meals of a single subadult sauropod individual,” says Poropat. “We don’t know if the plants preserved in our sauropod represent its typical diet or the diet of a stressed animal. We also don’t know how indicative the plants in the gut contents are of juvenile or adult sauropods, since ours is a subadult, and we don’t know how seasonality might have affected this sauropod’s diet.”

This research was supported by funding from the Australian Research Council.

“Despite major advances in monitoring, some volcanoes erupt with little or no detectable precursors, significantly increasing the risk to nearby populations,” said Dr Yuyu Li of the University of Illinois, lead author of the study in Frontiers in Earth Science. “Some of these volcanoes are located near major air routes or close to communities: examples include Popocatépetl and Colima in Mexico, Merapi in Indonesia, Galeras in Colombia, and Stromboli in Italy.

“Our work helps explain how this happens, by identifying the key internal conditions — such as low magma supply and warm host rock — that make eruptions stealthy.”

Warning signs

Veniaminof is an ice-clad volcano in the Aleutian Arc of Alaska. It’s carefully monitored, but only two of its 13 eruptions since 1993 have been preceded by enough signs to alert observing scientists. In fact, a 2021 eruption wasn’t caught until three days after it had started.

“Veniaminof is a case study in how a volcano can appear quiet while still being primed to erupt,” said Li. “It is one of the most active volcanoes in Alaska. In recent decades, it has produced several VEI 3 eruptions — moderate-sized explosive events that can send ash up to 15 km high, disrupt air traffic, and pose regional hazards to nearby communities and infrastructure — often without clear warning signs.”

To understand Veniaminof better, the scientists used monitoring data over three summer seasons immediately before the 2018 stealthy eruption, which produced only ambiguous warning signs immediately before it happened. They created a model of the volcano’s behavior in different conditions which would change the impact of a filling magma reservoir on the ground above: six potential volumes of magma reservoir, a range of magma flow rates and reservoir depths, and three shapes of reservoir. They then compared the models to the data to see which matched best, and which conditions produced eruptions, stealthy or otherwise.

Volcano by the numbers

They found that a high flow of magma into a chamber increases the deformation of the ground and the likelihood of an eruption. If magma is flowing quickly into a large chamber, an eruption may not occur, but if one does the ground will deform enough to warn scientists first. Similarly, a high flow of magma into a small chamber is likely to produce an eruption, but not a stealthy one. Stealthy eruptions become likely when a low flow of magma enters a relatively small chamber. Compared to observational data, the results suggest that Veniaminof has a small magma chamber and a low flow of magma.

The model also suggests that different conditions could produce different warning signs. Magma flowing into larger, flatter chambers may cause minimal earthquakes, while smaller, more elongated chambers may produce little deformation of the ground. But stealthy eruptions only happen when all the conditions are in place — the right magma flow and the right chamber size, shape, and depth.

However, when the scientists added temperature to their model, they found that if magma is consistently present over time so that the rock of the chamber is warm, size and shape matters less. If the rock is warm, it’s less likely to fail in ways that cause detectable earthquakes or deformation of the ground when magma flows into the chamber, increasing the likelihood of a stealthy eruption.

What next?

“To mitigate the impact of these potential surprise eruptions, we need to integrate high-precision instruments like borehole tiltmeters and strainmeters and fiber optic sensing, as well as newer approaches such as infrasound and gas emission monitoring,” said Li. “Machine learning has also shown promise in detecting subtle changes in volcanic behavior, especially in earthquake signal picking.”

At Veniaminof, taking measures to improve the coverage of satellite monitoring and adding tiltmeters and strainmeters could improve the rate of detection. In the meantime, scientists now know which volcanoes they need to watch most closely: volcanoes with small, warm reservoirs and slow magma flows.

“Combining these models with real-time observations represents a promising direction for improving volcano forecasting,” said Li. “In the future, this approach can enable improved monitoring for these stealthy systems, ultimately leading to more effective responses to protect nearby communities.”

To put the result in perspective: a person is more likely to be struck by lightning in a given year (1 in 1.2 million) than for one of Oxford’s quantum logic gates to make a mistake.

The findings, published in Physical Review Letters, are a major advance towards having robust and useful quantum computers.

“As far as we are aware, this is the most accurate qubit operation ever recorded anywhere in the world,” said Professor David Lucas, co-author on the paper, from the University of Oxford’s Department of Physics. “It is an important step toward building practical quantum computers that can tackle real-world problems.”

To perform useful calculations on a quantum computer, millions of operations will need to be run across many qubits. This means that if the error rate is too high, the final result of the calculation will be meaningless. Although error correction can be used to fix mistakes, this comes at the cost of requiring many more qubits. By reducing the error, the new method reduces the number of qubits required and consequently the cost and size of the quantum computer itself.

Co-lead author Molly Smith (Graduate Student, Department of Physics, University of Oxford), said: “By drastically reducing the chance of error, this work significantly reduces the infrastructure required for error correction, opening the way for future quantum computers to be smaller, faster, and more efficient. Precise control of qubits will also be useful for other quantum technologies such as clocks and quantum sensors.”

This unprecedented level of precision was achieved using a trapped calcium ion as the qubit (quantum bit). These are a natural choice to store quantum information due to their long lifetime and their robustness. Unlike the conventional approach, which uses lasers, the Oxford team controlled the quantum state of the calcium ions using electronic (microwave) signals.

This method offers greater stability than laser control and also has other benefits for building a practical quantum computer. For instance, electronic control is much cheaper and more robust than lasers, and easier to integrate in ion trapping chips. Furthermore, the experiment was conducted at room temperature and without magnetic shielding, thus simplifying the technical requirements for a working quantum computer.

The previous best single-qubit error rate, also achieved by the Oxford team, in 2014, was 1 in 1 million. The group’s expertise led to the launch of the spinout company Oxford Ionics in 2019, which has become an established leader in high-performance trapped-ion qubit platforms.

Whilst this record-breaking result marks a major milestone, the research team caution that it is part of a larger challenge. Quantum computing requires both single- and two-qubit gates to function together. Currently, two-qubit gates still have significantly higher error rates — around 1 in 2000 in the best demonstrations to date — so reducing these will be crucial to building fully fault-tolerant quantum machines.

The experiments were carried out at the University of Oxford’s Department of Physics by Molly Smith, Aaron Leu, Dr Mario Gely and Professor David Lucas, together with a visiting researcher, Dr Koichiro Miyanishi, from the University of Osaka’s Centre for Quantum Information and Quantum Biology.

The Oxford scientists are part of the UK Quantum Computing and Simulation (QCS) Hub, which was a part of the ongoing UK National Quantum Technologies Programme.