Interview'Each culture has an example of someone who speaks to animals'

TAU’s $10 million Doolittle prize seeks scientists who can ‘talk to the animals’

The award inaugurated this month at Tel Aviv University hopes to spur research, with the aim of creating human-animal dialogue ‘at their own level’

Gavriel Fiske is a reporter at The Times of Israel

Illustrative photo of a man training his dog, in the northern Israeli city of Tzfat, on January 10, 2019. (David Cohen/Flash90)
Illustrative photo of a man training his dog, in the northern Israeli city of Tzfat, on January 10, 2019. (David Cohen/Flash90)

When Tel Aviv University (TAU) first conceived of the $10 million prize, announced earlier this month, for demonstrating “true communication” with animals, one of the original ideas was to name the science contest after King Solomon, who in his famous, legendary wisdom was granted the ability to understand the languages of animals.

In the end, it was decided to name the “Coller-Dolittle Prize for Two-way Inter-species Communication” after English businessman and philanthropist Jeremy Coller, whose foundation put up the prize money, and the fictional character Dr. Dolittle, who got his start in a 1920s series of British children’s books and later became a beloved Hollywood icon.

Both King Solomon and Dr. Dolittle are depicted as conversing freely with animals and understanding them in much the same manner as they would other human beings — a common theme in fairytales and folklore but something that is actually impossible, said Coller-Dolittle Prize committee chair Prof. Yossi Yovel, of TAU’s zoology department.

“Each culture has this example of someone who speaks to animals… it’s a big dream [that] will never happen,” Yovel said, speaking to The Times of Israel by phone.

Instead, the contest aims to find ways to “communicate with animals at their own level, which is possible, although we are far from this,” he said.

The contest organizers offer specific criteria for such an achievement: Using “a non-invasive approach to communicate with or decipher an organism’s communication”; demonstrating communication in “more than one context (e.g., alarm, mating, foraging) using the organism’s endogenous communication signals”; and demonstrating a “measurable response of the organism to the signals broadcasted to it.”

Prof. Yossi Yovel of the Tel Aviv University Zoology Department. (courtesy TAU)

This means using advanced means such as artificial intelligence-based vocalization analysis and playback, look-alike robots to mimic movements, and other advanced technologies to “trick an animal into thinking it’s communicating with one of its own, in multiple contexts,” Yovel said.

A Turing Test for animals

It’s theoretically possible to replicate all the different ways animals communicate — vocalizations, olfaction, touch, vision and movement — but the less human the animals are, the more difficult it is to understand how to use technologies to communicate with them, even if they are simpler organisms, Yovel said.

Conversely, he said, animals that are closer biologically to humans, such as other mammals like dogs or monkeys, may have behaviors and communications that are easier for humans and computers to understand. At the same time, however, these are more complex organisms that present deeper challenges in communication.

“There is a solution with simple animals, it’s easier to manipulate them. For example with bees, we know their dances very well, maybe this is something robots could emulate. I would like to see how much of that is possible, or not,” Yovel said.

Flying bees swarm after collecting pollen in the spring. (Andreas Häuslbetz on iStock by Getty Images)

The contest idea is similar to the Turing Test, in which a computer AI model is measured by how much it can mimic human thought processes or “pass” as a human being during human-computer dialogue. In this case, the aim is to use a Turing Test model to enable a human to “pass” as an animal.

Using the bee example, that would mean advancing to the point where a human or computer-controlled robot bee could replicate a drone dance — which contains detailed information about where to find available pollen — well enough to convince a hive to set off in search according to the robot’s directions.

Communication, but no language

In general, animals are able to convey much more information than is usually assumed, and modern science is “understanding more and more about animal communication,” Yovel said.

He quickly gave several examples: It has been shown that dolphins have names, and different whale groups are known to have different kinds of songs. They can also learn songs from each other and broadcast them across great oceanic distances. Monkeys have alarm calls for specific predators, and some birds have been shown to be able to identify and keep track of their young through short, individualized calls.

Yovel noted a recently released study on elephants, the subject of a recent article in The New York Times where researchers, using AI to analyze elephant’s rich subsonic “rumbles,” appeared to show that elephants have individual names and would respond when those names were played back to them.

“AI knows how to receive a lot of data. It knows how to see patterns that sometimes humans can’t see. Animal perception can be totally different than ours. The problem is that AI can always find patterns, but we need to show that animals are responding to these patterns, we need to see that the animals also think similarly,” Yovel said.

Yovel’s own research is on bats, for which he maintains a lab at TAU. “I look at different aspects of their cognition, navigation, social behavior, parenting and vocalizations. With AI, we can take social calls from bats, we can see the context and understand if the bats are looking for food, or sex, or something else,” he said.

Illustrative photo of bats in the Teumim cave near Beit Shemesh. (Jorge Novominsky/Flash90)

Bats, like a lot of animals, “learn their communication. They aren’t born with a full vocal repertoire, they need to hear adult bats to learn and fine-tune it,” Yovel said.

It’s important to note that bat vocalizations, like those of other animals, aren’t a language like human languages. “It’s not the same. There are no real words and there is no syntax, there aren’t complicated statements.”

However, Yovel said, “Dogs can understand words, but can’t put them together.”

Humans have of course a long history of interacting and communicating with both domestic and wild animals, but Yovel said that training a dog to perform herding tasks or understand simple commands isn’t considered real communication, nor are instances of emotional bonding or friendships between humans and animals.

Even the famous examples of sign language using apes, which would seem to imply human-animal communication, are problematic and wouldn’t fit into the criterion of the contest, Yovel said.

“We taught them [to sign], it’s not theirs. Also, all these studies are very old and there is a lot of criticism. Depending on the context it’s hard to say how much the apes truly understood. Clearly [the researchers] succeeded, but it’s uncertain by how much,” Yovel said.

A goal of better treatment

Coller, who endowed the Coller School of Management at TAU, has long championed animal welfare through his Jeremy Coller Foundation, which has contributed extensively to research on the ills of factory farming by addressing “the consequences of intensive animal agriculture for human health, the environment, animal welfare, and global sustainability, with the aim of supporting the transition to a more sustainable food system,” according to the foundation website.

The Coller-Dolittle contest, sponsored by Coller’s foundation, is part of this effort, as it can spur research leading to “a better understanding of animals and how to treat them better in farms or zoos,” Yovel said.

The prize, open to research teams around the world, includes an annual $100,000 award for “contributions to decipher, interface or mimic non-human organism communication,” with the grand prize of $10 million in equity investments (or $500,000 cash) reserved for true animal communication, in which “the animal communicates independently without recognizing that it is communicating with humans.”

This year’s inaugural deadline for the annual contest is June 31, and the committee he leads has already received six or seven applicants, Yovel said. He declined to reveal details, but said, “We are open. We want to see interesting things.”

Most Popular
read more: