The Challenges of Animal Translation

Artificial intelligence may help us decode animalese. But how much will we really be able to understand?

Disney’s 2019 remake of its 1994 classic “The Lion King” was a box-office success, grossing more than one and a half-billion dollars. But it was also, in some ways, a failed experiment. The film’s photo-realistic, computer-generated animals spoke with the rich, complex voices of actors such as Donald Glover and Chiwetel Ejiofor—and many viewers found it hard to reconcile the complex intonations of those voices with the feline gazes on the screen. In giving such persuasively nonhuman animals human personalities and thoughts, the film created a kind of cognitive dissonance. It had been easier to imagine the interiority of the stylized beasts in the original film.

Disney’s filmmakers had stumbled onto an issue that has long fascinated philosophers and zoologists: the gap between animal minds and our own. The dream of bridging that divide, perhaps by speaking with and understanding animals, goes back to antiquity. Solomon was said to have possessed a ring that gave him the power to converse with beasts—a legend that furnished the title of the ethologist Konrad Lorenz’s pioneering book on animal psychology, “King Solomon’s Ring,” from 1949. Many animal lovers look upon the prospect of such communication with hope: they think that, if only we could converse with other creatures, we might be inspired to protect and conserve them properly. But others warn that, whenever we attempt to communicate with animals, we risk projecting our ideas and preconceptions onto them. We might do this simply through the act of translation: any human language constrains the repertoire of things that can be said, or perhaps even thought, for those using it.

In 1974, the philosopher Thomas Nagel published a seminal paper called “What Is It Like to Be a Bat?” Bat life, Nagel argued, is so profoundly different from human life that we can never truly know the answer to that question. Our understandings are shaped by our human concepts; the only way to know what it is like to be a bat is to be a bat, and to have bat concepts. Even if we don’t or can’t know exactly what it’s like to be a bat, we can have some understanding of how bat minds work; we can understand that bat life is lived aloft, sometimes upside down, and partly through echolocation. Still, in Nagel’s view, something is left out: the experience itself. As the philosopher Ludwig Wittgenstein famously put it, if a lion could talk, we could not understand him—our human minds would not share the sensory and conceptual landscape that lion-talk would express.

Today, animal-translation technologies are being developed that use the same “machine learning” approach that is applied to human languages in services such as Google Translate. These systems use neural networks to analyze vast numbers of example sentences, inferring from them general principles of grammar and usage, and then apply those patterns in order to translate sentences the system has never seen. Denise Herzing, the founder and research director of the nonprofit Wild Dolphin Project, which studies dolphins in the Atlantic, is now using similar algorithms, coupled with underwater keyboards and computers, to try to decode dolphin communications. “It may be that our mobile technology will be the same technology that helps us communicate with another species,” Herzing said, in a 2013 ted talk. An even more ambitious initiative called the Interspecies Internet—founded by the musician Peter Gabriel; the M.I.T. professor Neil Gershenfeld; the “father of the Internet,” Vint Cerf; and the cognitive psychologist and marine-mammal scientist Diana Reiss—seeks to use new technologies to connect intelligent species such as dolphins, elephants, and great apes to one another and to us. “Computer technology is finally allowing us to see inside the world of animals in ways that are showing us that they are complex sentient beings that deserve our understanding and respect,” Con Slobodchikoff, an animal behaviorist who is a professor emeritus at Northern Arizona University, said.

But other researchers, following Nagel, doubt that genuine translation can be easily achieved between species that don’t share the same basic perceptual and cognitive processes. “You can’t just have a Skype conversation,” Marcelo Magnasco, a physicist and dolphin researcher at the Rockefeller University in New York, told me. “We will need to understand what it is to be a dolphin.”

Not all animal minds are equally different. It stands to reason that mentally speaking, we have more in common with other primates than with octopuses and squid: the last common ancestor we share with chimpanzees lived six to eight million years ago, whereas the last we may share with octopuses lived in the Precambrian seas around six hundred million years before that. In “The Descent of Man,” Charles Darwin argued that there is “no fundamental difference between man and the higher mammals in their mental faculties,” and that the emotions of those animals were directly comparable to the ones that we experience. Darwin may have been indulging the Victorian tendency toward anthropomorphization, but it isn’t hard to discern a shared mentality with some of our animal cousins. Oliver Sacks expressed this memorably in his description of an encounter with an orangutan at the Toronto Zoo:

She stared into my eyes, and I into hers, like lovers gazing into each other’s eyes, with just the pane of glass between us. I put my left hand against the window, and she immediately put her right hand over mine. Their affinity was obvious—we could both see how similar they were. I found this astounding, wonderful; it gave me an intense feeling of kinship and closeness as I had never had before with any animal. . . . Then we pulled our faces away from the glass, and she went back to her baby.

“I have had and loved dogs and other animals,” Sacks wrote, “but I have never known such an instant, mutual recognition and sense of kinship as I had with this fellow primate.”

Read more here

Let's talk: What human language properties feature in animal communication? What is different about human language?