Nature

09
Apr

What Birdsong Can Teach Us About Creativity

Ofer Tchernichovski was waiting for a train to Manhattan when he heard a faint chirping. Looking around, he spotted a robin, fluttering on the platform. “His feathers were dull and he was all puffed up,” Tchernichovski recalls. “I don’t think he could fly. It wasn’t immediately apparent that he was singing—but he was.” The song was quiet, but elaborate. Given the bird’s condition, Tchernichovski thought that the bird might have been dying. But that didn’t seem to stop him from singing. “It was kind of touching. He was definitely focused on singing, even though the song was not directed at anyone.”

To most of us, birdsong is background, a part of nature’s soundtrack. But as a professor of psychology at Hunter College in New York City, Tchernichovski pays more attention than most. He slows down bird songs in an attempt to see what kinds of vocal tricks are up their sleeves. He prods birds to see if they’ll deviate from traditional tunes. Then he tries to put himself in the bird’s head. “We’re trying to understand these songs more intimately,” he says. “What about the song is getting them in a sexy mood?”

singing-robin
A robin's metallic warbling is just one example of a complex bird song.

Scientists began researching birdsong more than a century ago, and everyone pretty much agrees that birds use it as a sexual selector—that is, to find a mate or to protect their territory. But there may be more to it than simple reproduction. “I haven’t heard a reasonable explanation of why birdsong is so complicated,” says Jeff Markowitz, a graduate student in computational neuroscience at Boston University.

To us, birds’ singing isn’t unusual. But the ability to sing is actually quite rare in evolutionary terms. Plenty of animals can make sounds, but few in the same way birds do. What’s more, only a small subset of animals is capable of learning vocalizations when they’re young by imitating other members of their species. That puts songbirds in very rarified territory along with other big-brained animals, including humans, dolphins, and elephants. In fact, structured vocalizations are so fundamental to being a songbird that they have a brain area called the high vocal center that is almost entirely dedicated to singing.

“That’s not true of brain parts, generally. They usually do a bazillion different things,” says Elizabeth Regan, professor of psychology and neurobiology at Cornell University. That dedicated region of the brain not only makes songbirds a curiosity in the animal kingdom, it also provides scientists with a unique opportunity to study how groups of neurons collaborate to produce melodic sounds. “Birdsong is a great example of a behavior that’s interesting and complicated, and one where you might be able to figure out how the brain is actually producing it.”

That’s why Tchernichovski, Markowitz, and a number of other researchers are working to unravel why birds sing and the cognitive processes that underpin it. Along the way, they’re trying to answer another, deeper question: Is there some element of creativity in birdsong—some ability to think flexibly, to break from the script and rewrite an age-old tune? Or is birds’ apparent artistry just an accident of biology?

Crafting versus Composing

Birdsong isn’t the only evidence we have that some birds are preternaturally intelligent. Many pet owners have parrots that are clever escape artists, deftly unlatching the door to their cages. Ornithologists and psychologists have taught trained ravens how to master multistep procedures to obtain a treat. And other wild corvids have been observed bending sticks into spears—using tools, in other words. But these examples are all physical puzzles, more akin to building a Lego set or assembling some Ikea furniture. Birdsong is something else entirely. It’s a structured form of expression, similar to writing a poem or, more directly, composing a song.

This excerpt from "Inside Animal Minds: Bird Genius" showcases Muppet, a cockatoo with superior lock-picking skills.

For many birds, singing means following a script, the successful reading of which results in mating. Tunes are handed down generation to generation—most species have a characteristic song that they learn from their fathers early in life. Like infant humans, young birds imitate sounds when their synapses are still malleable, what scientists call a period of high neuroplasticity or, more colloquially, a “babbling stage.”

This period of exploration is like a child learning to play the piano—she will explore and press down on the keys randomly until she picks up on scales and other musical conventions from her teacher. Unlike humans, once most birds reach sexual maturity, song patterns are cemented in place. But that doesn’t mean they’re all the same.

Learning to Combine

The study of birdsong structure is a delicate balancing act. Scientists need to analyze simple songs in controlled environments before moving on to more complicated scenarios. The domesticated zebra finch makes for a perfect starting point—its song lasts for only a couple seconds, and you only need to monitor a few neurons to get a reasonably accurate portrait of what its brain is doing.

That’s why when Tchernichovski and his colleagues from New York City, Israel, and Japan wanted to understand how birds combine and arrange the units of sounds known as “syllables” during their youth, they turned to zebra and Bengalese finches, another favorite species. The scientists trained young birds to do two things—incorporate new syllables into their songs and swap others around. To do so, they played a recording of a song that’s outside of the birds’ repertoires and tallied each bird’s progress as it learned the new tune. Researchers documented the syllabic order of the transitions and how many times the birds repeated each part of a song before they could master a new one.

zebra-finch2
Scientists study zebra finches to understand the basics of vocal learning in birds.

Tchernichovski and his colleagues found that all of the birds completed both tasks in steps. For example, if a bird is taught to switch its song from ABC-ABC to ACB-ACB, it starts by learning one transition, say CB, by repeatedly singing ABC-BCBC, often several thousand times over a period of days. Then it learns another transition, AC. Once the bird learns the third transition BA, it’s finally able to sing the new song ACB-ACB from start to finish.

The finches’ stepwise learning pattern is sort of like learning how to play solitaire or chess. To get your cards or pieces arranged the way you want them, you might need to make two different moves. To do that, you may have to master one move first before doing both together. Speech in infant humans develops through the same process, too—Tchernichovski’s team found that for babies, the process of adding a new syllable lasts up to 20 or 30 weeks. But unlike zebra finches, we continue to rhapsodize and explore our entire lives.

A Canary Song’s Complexity

Past puberty, all birds—even if they one know just one song—do have some amount of freedom, but some have more than others. The Bengalese finch, which has a vocal ability that’s a bit more complex than the zebra finch, is a good example: it can loop back to preceding syllables at various stages of its song, like a DJ replaying a part of a tune. It’s an added an element of complexity that zebra finches can’t match, one that could be a glimmer of creativity. But it could also just be a product of chance.

“Generating complexity might turn out to be a less elaborate process than we think,” Markowitz says. In fact, researchers are discovering that it’s surprisingly simple to develop algorithms that can produce the same sort of complex patterns that we see in some bird’s songs.

The songs of other species, though, don’t boil down to an algorithm so easily. There may be unexpected breaks or surprising structures. Some birds may even actively compose their songs. For example, nightingales can pause mid-song for a few seconds and then pick up exactly where they left off rather than starting over again—a sign that they are considering their past chirps when deciding what to tweet next. And by statistically analyzing the songs of western meadowlarks and American redstarts, researchers have found evidence that after a bird has sung in a certain style, it’s less likely to repeat that until it takes a break. It’s as though the bird remembers its past performances and is trying to change things up.

Canaries, whose complex songs contain extremely flexible phrases, have traces of this second, more nuanced kind of memory. Markowitz and his team wanted to test the Belgian waterslager canary’s “memory” by statistically analyzing their songs. They looked at 34,000 phrases and over 100 songs to see how likely birds were to tweet each sequence and discovered that what a canary had sung for the previous 10 seconds seemed to influence what it was singing at the present moment.

Other studies in recent years only found evidence of bird song memories that stretch back just a tenth of a second. For you and me, that would be like remembering only the last word we said. But canaries have more of their song in mind at any given moment. It’s similar to thinking in paragraphs rather than individual words. And while the high-probability sequences Markowitz teased out could be rote, everyday singing, the low-probability sequences—the rare ones that don’t get sung a lot—could be evidence of creativity. The birds could be riffing on what they already know and violating musical expectations to come up with something new.

It’s a compelling idea, but it’s hard to pin down as true creativity without knowing what’s going on inside the canary’s brain. Until we can match birdsong probabilities with actual bird brain patterns, we won’t know if what we’re seeing is artistic or accidental.

“We don’t know if the canary is making a decision and then, in some sense, remembering that decision and having it impact their song 10 seconds later,” Markowitz says. “We want to know how these models actually match the reality of songbird neurobiology—and that we simply don’t know.”

Creativity or Mere Chemistry?

In an attempt to find out, Markowitz is proposing a mashup, a side-by-side blending of birdsong statistical models and active brain patterns. That should help determine whether these birds are actually being creative or whether they’re just obeying a complex set of biologically-enforced rules. Doing so could help scientists understand the underlying biology of birdsong.

For example, his mashup could build on a theory known as “sexy syllables,” first proposed 15 years ago. Then, animal behavior specialists discovered that when male canaries sang a brief, two-note syllable, female canaries perked up their tails and ruffled their feathers in solicitous displays.

Perhaps these syllables, Markowitz says, leads to a dopamine rush in the male’s brain. That rush could then prompt him to select a different syllable or flourish 10 seconds later that’s designed to surprise the female yet again. What looks like conscious decision-making on the outside may just be the bird responding to some basic physiological rules. Until we can see into the bird’s brain while it’s singing in a social environment—how the female responds at a given time, how other competing birds’ songs factor into the probabilities—we can’t actually know.

yellow-canary
Canary songs are significantly more elaborate than those of zebra finches, for example.

Scientists may also find unnecessary components of canary song—ornaments that don’t help or perhaps even hinder a male bird in his quest to find a mate. By removing parts of the song that researchers suspect aren’t important, they could see how the females behave in response. If females react the same to a simplified song as they do to a full song, it’s possible that the additional notes are the male canary just playing around. “I think with the right tools, you can start to look at birds like the nightingale, the brown thrasher, and other wild, complex singers,” says Markowitz. “We may find that all of this complexity serves absolutely no purpose.”

So if birds can strip a song down to a few naked notes and still achieve the same result, why would these embellishments have evolved in the first place? “It might be arbitrary,” Regan explains. “Maybe the females just want something new and different—they want something fancy,” Regan says. Or maybe they have been selected as a signal of fitness, she adds. “You can’t have a terrific brain unless you’re in good health, had a good upbringing, and have a good genome.”

Channeling Our Inner Bird

Even if songbirds’ flourishes ultimately have no discernable purpose, they are still helping scientists to pin down the role of variability in the learning process. “To learn something very robustly, you have to explore,” Markowitz says. That means walking through different possibilities, learning new sounds, singing duets with other birds like wrens do.

These discoveries are also highlighting other similarities between the way humans and birds learn. Tchernichovski and his team discovered, for example, that finches go through a sort of recovery period when they first wake up. The quality of their song isn’t as good at first, but after a short time, it returns to normal. If the bird doesn’t get a good night’s sleep, though, that recovery period is longer and their performances suffer. But if they do get a good night’s sleep, their songs can be even more elaborate at the end of the recovery period.

“What the bird is doing each morning is going back to a more plastic state,” Tchernichovski says. After sleep, their brains are more elastic, allowing for more experimentation. Then as the day wears on, the screws in their brains tighten up and their songs snap into place. The next morning, though, those structures have loosened again and the brain has returned to its more flexible state. “We found that the birds who oscillate in this way—more structure, less structure, more structure—learn better.”

Given the complexity of birdsong and how birds go about it, some scientists conclude that it’s a mistake to think that songbirds sing for purely sexual reasons. Tchernichovski is one. He recalls gazing upon the dying robin at the train station with a mix of sadness and sanguine curiosity. At that moment, it certainly wasn’t looking for a mate—and yet it was still singing. “This almost makes me feel better,” Tchernichovski says. “It’s as if the bird is somehow comforting itself. It seems to be more encouraged, even in sickness, by singing.”

Like humans, birds seem to put a bit of personality into each performance. And that, scientists are discovering, requires some intelligence. “Seeing that kind of complicated behavior acted out by such a simple species definitely makes me think that, from one perspective, what they’re doing is not human,” Markowitz says. “But from another perspective, what we do is a little more like what birds do than we might think.”