Latest news with #YossiYovel
CBC
2 days ago
- Science
- CBC
Like real-life Dr. Dolittles, scientists are using AI to decode animal communication
Bottlenose dolphins are known for their intelligence, and now researchers are trying to find out whether we could one day communicate with them in their own language. Researchers from Woods Hole Oceanographic Institution (WHOI) in Massachusetts and the Sarasota Dolphin Research Program in Florida are using artificial intelligence to decode the meaning behind dolphin whistles. "Our objective is to understand their rules of communication, what the structure, function, and meaning of dolphin communication is," Frants Havmann Jensen, an investigator at WHOI's Marine Research Facility, told The Current's Matt Galloway. "So, not just identifying the sounds they make but uncovering what those sounds mean to them." In May, the researchers were awarded the Coller Dolittle Challenge for Interspecies Two-Way Communication for that work. It honours researchers who've made significant scientific advances that could pave the way for human-animal communication. Yossi Yovel, who led the judging panel for the Coller Dolittle prize, says the Jeremy Coller Foundation is interested in unlocking a deeper understanding of language, across species. "By understanding how communication has evolved across many different species, we can better understand the evolutionary roots of communication and language," he said. Yovel says understanding the signals and the messages they convey is a crucial first step to decoding bottlenose dolphins' communication system. From there, scientists can begin to understand how dolphins organize signals when they're communicating to create what humans would understand as sentences. "The next step would be to present signals that you've discovered to the animal and observe their response, and to show that you can do this in multiple contexts," he said. Using AI to enhance understanding The Sarasota Dolphin Research Program is conducting the world's longest-running study of a wild dolphin population. Since 1970, they've built a database of sounds from over 300 dolphins. Jensen says bottlenose dolphins have distinct, individual sounds researchers call signature whistles. "It's the dolphin equivalent of a human name. Dolphins use these signature whistles to maintain social bonds and recognize each other," he said. Dolphins also make non-signature whistles, which comprise approximately 50 per cent of the whistles they produce, but there's little research in this area. The study published by the winning team suggests that the non-signature whistles could function like words with mutually understood, context-specific meanings. Jensen says AI can help researchers decode the dolphins's communication by automatically detecting and discovering new shared whistle types. "We're looking into how to use it for identifying patterns of use across individuals and contexts so that we can begin to infer meaning from how dolphins use these," he said. Jensen and other researchers say one of AI's strengths is its ability to process large amounts of data. Sophie Cohen-Bodénès and her team at Washington University in St. Louis, Mo., — who were shortlisted for the prize — are using AI to decipher patterns in cuttlefish arm wave signals, a form of sign language. Through non-invasive behavioural experiments, Cohen-Bodénès examined that the creatures interpret arm signs using vision and vibrations. "We're in the process of collecting large datasets from many behavioural contexts to give to the AI algorithm that could find, in an objective way, the different correlations between different arm signs," she said. Cohen-Bodénès says her research goal is to gain more insight into the meaning of animal communication displays and their underlying sensory mechanisms. "It's a way to better assess their welfare, to better understand their needs and to improve their protection." WATCH | Dolphins circle space capsule: #TheMoment dolphins greeted the capsule returning astronauts to Earth 3 months ago Duration 1:04 Marine mammal expert Ashley Noseworthy recounts the moment a pod of dolphins greeted the SpaceX capsule carrying NASA astronauts returning from nine months stuck in space. Limits of AI Yovel says AI is a powerful tool but it has shortcomings. When researchers consider the meaning and context of their findings, he says they base them on human observations, which could be limited or wrong. Yovel believes humans might be able to communicate with animals, but he's skeptical that AI could be used in a device or algorithm that would allow them to have a conversation. Human communication is complicated and allows people to discuss a wide range of topics, and Yovel doubts animal communication systems are as complex. "We have this language, which is extremely complex, and it seems to stand out in comparison to other animal communication systems," he said. In order to understand whether human language systems stand out and how, Yovel says humans need a better understanding of nature.
Yahoo
31-03-2025
- Science
- Yahoo
How bats avoid crashing into one another
When a colony of bats leaves their cave and takes to the skies at night to hunt, they often do so in such big groups that they almost look like one giant blob. How these winged mammals can achieve such tight densities–sometimes numbering in the hundreds of thousands of bats–yet do not crash into one another has puzzled scientists for decades. Now, researchers believe they've figured out how bats can still hear amidst the din similar to a noisy cocktail party. They appear to change the way they echolocate in order to get a better idea of where exactly the bats closest to them are located. The findings are detailed in a study published March 31 in the journal Proceedings of the National Academy of Sciences. Like dolphins, toothed whales, and some birds, many bat species use echolocation to perceive the world around them. They send out a call and listen for the reflected echo. This echo allows them to 'see' what is in front of them. However, if many bats are echolocating at once—like when an entire colony comes out of a cave in only a few minutes—the noise from others should cover up the critical echoic information that bats need to get around. This loss of acoustic information is referred to as 'jamming.' Bats should collide because of the jamming, but aerial accidents outside of caves are actually quite rare. 'You're almost excited when you witness one,' Aya Goldshtein, a study co-author and a postdoctoral scholar in animal behavior at the Max Planck Institute of Animal Behavior in Germany, said in a statement. Video showing the evening emergence of thousands of Greater mouse-tailed bats, as they take to the sky in search of insects. The video shows rare collisions of bats in mid-air. CREDIT: Yossi Yovel and Eran Amichai. Scientists have deployed several techniques in the lab to try and figure out how bats can hear around all of this ambient chatter–like the noise inside of a crowded restaurant. They've studied how bats echolocate in groups, by seeing how the bats ecolocated at a slightly different frequency. In theory, this should reduce jamming, but the results from previous studies did not provide a strong enough answer as to how or why. 'No one had looked at this situation from the point of view of an individual bat during emergence. How can we understand a behavior if we don't study it in action?' Yossi Yovel, a study co-author and neuroecologist from Tel-Aviv University in Israel, said in a statement. For this study, the team studied Greater mouse-tailed bats (Rhinopoma microphyllum) living in Israel's Hula Valley and collected data directly from wild bats emerging from a cave at dusk over two years. They tagged several bats in this colony with lightweight trackers that recorded their location every second. Some of the tags were also equipped with ultrasonic microphones that recorded the soundscape from a bat's point of view. To study this behavior, they used a new combination of high-resolution tracking, ultrasonic reporting, and sensorimotor computer modeling to observe how the bats' sense as they squeezed out of the cave opening. However, the tagged bats were released outside the cave and into the emerging colony. meaning that that data taken from the cave opening when the density is highest was missing. To fill this gap, the team used a computational model developed by study co-author Omer Mazar, which simulated emergence. This model used data collected by the trackers and microphones to recreate the full behavioral sequence beginning with the entrance of the cave and wrapping up after the bats had flown 1.2 miles (two kilometers) through the valley. 'The simulation allows us to verify our assumptions of how bats solve this complex task during emergence,' Mazar, a Ph.D student at Tel-Aviv University, said in a statement. The data revealed that when exiting the cave, the bats experience a cacophony of calls. Ninety-four percent of echolocations are jammed, yet the bats significantly reduced the echolocation jamming within five seconds of leaving the cave. The bats were also observed making two important behavioral changes. First, they fanned out from the dense colony core while keeping the group structure. Next, they sent out shorter and weaker calls at higher frequency. Initially, the team suspected that the bats would reduce the jamming by quickly dispersing from the cave. Instead, they appear to change their echolocation to a higher frequency, even though in theory that should only increase the jamming and collision risk. To understand why, the team had to approach this whole scene from a bat's point of view. [ Related: What bats and metal vocalists have in common. ] 'Imagine you're a bat flying through a cluttered space. The most important object you need to know about is the bat directly in front. So you should echolocate in such a way that gives you the most detailed information about only that bat,' explained Mazar. 'Sure, you might miss most of the information available because of jamming, but it doesn't matter because you only need enough detail to avoid crashing into that bat.' It appears that the bats change the way they echolocate in order to gain detailed information about their neighbors nearby. This strategy ultimately appears to help them successfully maneuver and avoid bumping into one another. According to the team, this unexpected result for how bats solve this noisy dilemma was made possible by studying bats in their natural environment. 'Theoretical and lab studies of the past have allowed us to imagine the possibilities,' said Goldshtein. 'But only by putting ourselves, as close as possible, into the shoes of an animal will we ever be able to understand the challenges they face and what they do to solve them.'
