Before toddlers learn to master symbolic counting–where “one” stands for a single object and “two” indicates double that–they pass through a developmental stage of verbal tallying. In this phase, asked how many apples are in a group of three, a young child might say “one, one, one” or “one, two, three” or “apple, apple, apple,” in a form of proto-counting. They understand there are three apples, but they don’t quite yet have the ability to express that number in the abstract form of “three,” alone.
Inspired by this knowledge, Diana Liao, a neurobiologist and post-doctoral researcher at the University of Tubingen in Germany, decided to ask the obvious question: Can crows do it too? Spoiler alert: they can, according to a first-of-its-kind study published May 23 in the journal Science. Carrion crows can control their vocalizations and correspond the number of those calls to a cue, in a form of proto-counting, per the new research.
The study adds to the growing laundry list of cognitive abilities that corvids (the bird family including crows, ravens, magpies, and jays) possess. The new work also aids in the quest to uncover the evolutionary origins of humans’ mathematical ability. By studying other animals’ capacities and limitations, scientists can get a better idea of where and how our own numeracy comes from.
“It’s a really neat study,” Jessica Cantlon, a professor of psychology at Carnegie Mellon University who researches quantitative reasoning and was uninvolved in the crow investigation, tells Popular Science. In humans, the connection between the vocal system and our mathematical brain is a key part of our number sense. “When you start using vocalizations to represent things out in the world it kind of changes the way you think…that [is the way that] counting emerges in humans over development and over evolutionary time,” she explains. “It hasn’t been clear that any other animal could do that,” she Cantlon adds–until now.
Previous research–usually involving pecks, button pushing, food choice, clicks on a screen, head movements, or hand-singing–has demonstrated that crows, chimpanzees, monkeys, parrots, rats, honeybees, chickens, and other animals do have some level of quantitative reasoning–including distinguishing between different amounts of things, conducting basic arithmetic, even, rarely, conceptualizing zero. But humans’ level of vocal control is unique among animals. Even other primates “are so bad at producing human-language-type vocalizations,” says Cantlon. “I think people kind of gave up trying to get primates to [vocalize] in quantitative reasoning tasks.”
Crows are just one entry on the long list of species that make sounds, adding to the “ca-caw”phony of the natural world. In contrast to people though, scientists think most animal calls are involuntary and instinctual–programmed in at birth and elicited by external environmental cues, Liao explains. Research on chickadees, for instance, has shown that the birds will add more “dee” syllables to their alarm calls, in the presence of larger predators–but that trend doesn’t necessarily signal that the songbirds are literally measuring predator size with their tweets. Instead, it’s possible that the number of “dees” corresponds to the amount of arousal elicited by the presence of a predator, and larger threats are more alarming. Liao wanted to dig into this phenomenon in a more controlled setting, where an artificial lab set-up would enable her and her colleagues to eliminate the role of instinctual fear and arousal.
The new study of carrion crows is the first to explicitly show that any other species besides humans can link voluntary vocalizations with an understanding of quantity. Keeping tabs of caws “is not like counting the number of pecks,” says Girogio Vallortigara, a neuroscientist at the University of Trento in Italy who has published on numeracy in zebra fish and newborn chicks. Instead, tracking vocalizations involves “timing…and requires extensive memory and planning,” he explains. “It is certainly an advanced cognitive ability.”
In order to prove what the corvids were capable of, Liao and her co-authors trained three carrion crows to associate colored arabic numerals (1, 2, 3, and 4) and neutral auditory cues (i.e. sounds that weren’t related to any natural threat or prey) with a corresponding, flexible series of vocalizations. The process took over a year, and mostly unfolded through trial and error, Liao says. The researchers would present each crow with a number on a screen or a pre-recorded sound and then reward the bird with food if and when it produced the matching number of calls.
Once training was complete, the crows were tested in repeated one to two hour sessions. Each bird, positioned in front of a screen, initiated trials by pecking a target. Then, a number would appear on the screen in response, or an audio cue would play. The crows would call in response, and then peck an “enter key” on the screen to indicate the end of their response. If the number of calls correctly corresponded to the number cue, the crows would get a tasty mealworm or birdseed pellets as a prize.
Across 20 sessions each, all three crows demonstrated they could line up their vocalizations with the cues at a much higher rate than chance. Most of the time, the crows got it right–displaying 100% accuracy in response to the number one, more than 60% accuracy for number two, over 50% for three, and about 40% accuracy for the number four. Using a computer analysis tool, the researchers further found that the first calls in a series differ significantly, depending on the number of calls to follow–indicating that crows are planning their response and call number from the very first caw. Finally, the researchers assessed the trails where crows got it wrong, and found that errors fell into two categories: stutters (where birds repeat a call) and skips (where a bird forgets one)–suggesting that when the birds get it wrong, they generally start with the correct plan in mind and then sometimes lose track along the way.
In total, their experiments suggest carrion crows have an advanced capacity to make or withhold vocalizations in response to arbitrary cues–tallying aloud as toddlers do.
There are some limitations to the study, however. The authors didn’t measure arousal directly, so it could potentially still be a factor in the birds’ response, says Liao. And it’s possible that the birds were tracking call duration, not discreet syllables, in their responses, says Cantlon. If that’s the case, they may not actually have the ability to understand one-to-one associations between a call and a number of things, it could just be that they associate a length of time with a stimulus. Further research would be needed to know for sure.
Liao plans to conduct additional experiments exploring how crows might use vocalization numbers in the wild, and also hopes to examine the underlying brain systems supporting the newly discovered ability. “The avian lineage diverged from the primate lineage over 300 million years ago. We’ve evolved dramatically different brain architectures. It would be so interesting to see how different brains come up with similar behaviors.”
Down the road, what we learn about crows could help us better understand ourselves. Mathematics is a “defining characteristic” of human cognition, but “it’s not as if we’ve invented ways of thinking that are completely unique,” says Cantlon. “Everything about being human has a backstory, and that’s something we can observe in modern species. If we look at different branches of the evolutionary tree, we can start to trace the history of our own thinking.”