Some 40 kilometers east of the Tanzanian coast in East Africa lies Latham Island, a rocky, utterly isolated and uninhabited piece of land about the size of seven soccer fields. It was on this unlikely patch of ground that Weizmann Institute of Science researchers recorded – for the first time ever – the neural activity of mammals in the wild. In their study, published today in Science, the team used a tiny device to record, at the level of single neurons, the brain activity of fruit bats as they flew around the island. The scientists discovered that the bats’ neuronal “compass” is global: It provides stable directional information across the entire island and does not depend on the Moon or stars. Many species share the behavioral ability to orient themselves using an “internal compass,” and it is quite possible that humans rely on the same neural mechanism that was studied in these bats.

In 2018, Prof. Nachum Ulanovsky of Weizmann’s Brain Sciences Department embarked on a worldwide search for a natural setting that would allow him to study mammalian navigation in the wild. “I was looking for an area that was large enough to release bats and follow how they navigate, but not too large, with no tall trees and isolated from other land, so that we could easily recapture the bats and recover the recordings of their brain activity,” Ulanovsky explains. “You might think there are countless suitable islands, but even after a systematic worldwide search, we couldn’t find the right one. Night after night, I inched the cursor across Google Earth, searching for an island in the middle of the ocean. One night, I zoomed in on a region I had scanned before – and suddenly discovered Latham Island.”

What does a neuroscientist take to a desert island? Ulanovsky and his team brought camping gear, satellite communication equipment and a great deal of scientific apparatus, all shipped from Israel to Tanzania. They hired local fishermen to provide food and ferry them back and forth to the island. “The island is near the equator; during the year there are two dry seasons with generally pleasant weather,” he says. “We launched our expedition in February 2023, when weather conditions were expected to be comfortable. After renting a building at Tanzania’s central veterinary institute, we renovated it and set up a laboratory. We selected six local fruit bats of the same species we had previously studied in Israel and implanted in them tiny devices that record brain activity and transmit their location using GPS. This is the smallest device of its kind in the world, developed specifically for this study. Then we sailed to the island. Unfortunately, Cyclone Freddy, the longest-lasting tropical cyclone ever recorded, was still raging about 1,500 kilometers to the south, generating strong winds on the island and preventing the bats from flying during the first week. Eventually the weather cleared, and we began the experiment. On our second trip, in 2024, the weather was much kinder and we encountered no storms.”

The expedition, led by Shaked Palgi, Dr. Saikat Ray and Dr. Shir Maimon from Ulanovsky’s lab, first allowed the bats to acclimate to their new surroundings in a flight tent. Afterwards, each bat was released to fly alone for 30 to 50 minutes every night. While the bats flew, the researchers recorded the activity of more than 400 neurons deep in their brains, in regions known to be involved in navigation. They found that every time the bats flew with their heads pointing in a particular direction – north, for instance – a unique group of neurons became active, creating an “internal compass.” Navigation by means of directional neurons had previously been observed in the lab, but this was the first evidence that it happens in nature as well. When the researchers analyzed the recordings from different parts of the island, they discovered that the activity of the head-direction cells was consistent and reliable across the entire island, enabling the bats to orient themselves over a large geographical area.

“One of the big questions in mammalian navigation is whether head-direction cells function as a local compass or as a global one,” Ulanovsky explains. “In other words, does a given group of cells always point in the same direction – north, say – or does the entire compass reorient itself depending on the local environment? We found that the compass is global and uniform: No matter where the bat is on the island and no matter what it sees, specific cells always point in the same direction – north stays north and south stays south. We also saw that when a bat moved from the western coast of the island to the southern coast, the change in coastline direction did not disrupt the compass. Additionally, the compass remained true even when the bats flew at different speeds and altitudes.”

The next question was what information the bats’ compass relies on. We know that many migratory birds use the Earth’s magnetic field, whose direction is uniform, just like a human-made compass. However, this did not appear to be the case with bats. “During their first nights on the island, the neuronal compass activity was not too stable,” Ulanovsky says. “We observed a gradual learning process until, by the third night, the bats’ compass orientation became very stable. Such learning doesn’t fit with using the magnetic field, which had been there since the very first night.”

Another way of orienting oneself in space is by relying on landmarks in the environment, such as tall buildings in a big city. “Our findings suggest this is the most likely possibility, and it fits with the need to get to know a new environment over several days,” Ulanovsky says. “Every natural environment is full of landmarks that can be seen, smelled or heard. Latham Island’s topography included cliffs and large boulders that could serve as navigation cues. In fruit bats, sight is the dominant sense and has the longest range, so we assume they mainly rely on vision. Unlike navigation based on magnetic fields, a system that depends on learning landmarks requires complex neural computations, partly because only some of the landmarks are visible from any given point. That’s precisely why using this system takes several days of learning – or in fact, several nights.”

Could bats, like humans and other animals, also look upward and navigate using the Sun, the Moon and the stars? Celestial bodies are unstable cues – they appear, move and then disappear – so using them for navigation is complicated. Laboratory studies had previously shown that moving objects resembling celestial bodies can affect the activity of head-direction cells in mammalian brains, but when the researchers recorded the activity of these cells in bats flying in the wild before and after moonrise, they detected no change. Likewise, the bats’ internal compass remained stable and accurate regardless of whether the Moon and stars were visible or hidden by clouds.

“We found that the Moon and stars are not essential for bats to navigate,” says Ulanovsky. “Still, it’s possible that their compass integrates celestial cues with local landmarks. The angle of celestial bodies relative to an animal does not depend on its exact location, so these bodies can serve to calibrate the compass. For example, on their first night in a new environment like Latham Island, bats could compare the position of landmarks with that of celestial bodies supplying an ‘absolute truth’ – which would greatly speed up learning and stabilize the compass.”

Head-direction cells are the most basic navigational mechanism in mammals, appearing at the earliest stage of brain development after birth. They are also evolutionarily conserved, found in species ranging from flies to rodents to bats. “Until recently, a person unable to navigate would not have survived,” Ulanovsky says. “Even today, being able to orient oneself can be a lifesaver. Studying mammalian navigation helps us hypothesize how navigation mechanisms work in the human brain and how they can become disrupted, for instance in neurodegenerative diseases such as Alzheimer’s. Although our lab at the Weizmann Institute offers conditions that simulate natural environments, including large flight rooms and a 200-meter-long bat tunnel, even these setups lack the full complexity of the wild. Until recently, studying brain activity in natural conditions was impossible; our ability to finally do so now is due in part to technological advances and miniaturization.”

“Of course, conducting research in the wild is complex and unpredictable,” Ulanovsky adds. “For example, we had to ask a commercial satellite company to shift its satellite slightly so we could get reception on the island. Despite the challenges, our findings show there is no substitute for testing lab-based knowledge in the real world. We hope our study will encourage other groups, in brain sciences and beyond, to take their research out of the lab and into nature.”