The Elephantine Memories of Food-Caching Birds

Black-capped Chickadee. Photo: Donna kenski

by Matthew Hutson

reprinted with permission from The New Yorker

Awhile ago, I searched for a beard trimmer in my bedroom. I spent probably forty-five minutes looking in every likely location at least twice, and every unlikely location at least once. I swore up a storm; the trimmer never turned up. I’ve played similar games with pants. There’s a reason for the burgeoning market in electronic tags that track your belongings.

Our poor memories can seem mystifying, especially when you consider animals. This time of year, many species collect and cache food to stave off winter starvation, sometimes from pilfering competitors. So-called larder hoarders typically keep their troves in a single location: last year, a California exterminator found seven hundred pounds of acorns in a client’s wall deposited there by woodpeckers. In contrast, scatter hoarders—including some chickadees, jays, tits, titmice, nuthatches, and nutcrackers—distribute what they gather over a wide area. Grey squirrels use smell to help them find their buried acorns. But many scatter hoarders rely largely on spatial memory.

People first noticed scatter hoarding by 1720 or even earlier. It’s come under serious investigation, however, only in the past century. Scientists now know that birds’ brains can contain elephantine powers of recollection. Some birds can store, or cache, tens or even hundreds of thousands of morsels in trees, or in or on the ground, and retrieve a good portion of them. In 1951, a Swedish ornithologist named P. O. Swanberg reported on Eurasian nutcrackers: over the course of a single autumn, he saw each bird make some eight thousand caches. That winter, the birds dug through the snow to retrieve their stored food. Swanberg examined the excavation holes the birds had left behind and found nutshells in nearly ninety per cent of them—an indication that there had been few fruitless efforts.

In the late nineteen-seventies, researchers at Oxford buried sunflower seeds just ten centimetres from where marsh tits had buried their own morsels. Over the following days, the bird-buried seeds disappeared significantly faster than those the scientists had buried, suggesting that the birds had precise memories of their cache locations. In 1992, other scientists reported that birds known as Clark’s nutcrackers could recall, with better-than-chance accuracy, where they’d buried seeds more than nine months earlier.

Vladimir Pravosudov began studying food caching as an undergraduate at Leningrad (now St. Petersburg) State University. “I’m a big believer in just watching animals,” he told me. Above the Arctic Circle, he’d spend hours a day with binoculars and a stopwatch, observing willow and Siberian tits; he found that they could cache food as frequently as twice a minute. Extrapolating, he estimated that they could store as many as half a million bits of food each year. He grew fascinated with the question of how and why birds had evolved to be “caching machines.”

Today, as a biologist at the University of Nevada, Reno, Pravosudov focusses on chickadees. It’s impossible to measure a wild bird’s retrieval accuracy precisely: among other things, there’s no way of knowing when a bird looks in the wrong place before the right one. Still, “every time I see them recovering, they look very purposeful,” Pravosudov said. When a chickadee retrieves a seed from bark or lichen, he noted, “They don’t search, they just go boom, and they just pull it out. When you see this, it’s very, very impressive.”

All of which raises a question: If a bird can remember where it’s placed thousands of seeds in the forest, why can’t I find my pants? It would be good to know how birds evolved such extraordinary memories so different from our own.

To find out how natural selection may have forged birds’ brains into steel traps, Pravosudov uses a unique piece of equipment called a feeder array, which his lab installs in the Sierra Nevada mountains. Each array comprises a square frame, about four feet per side, which holds eight automated feeders. Pravosudov’s team catches chickadees in nets and straps tiny electronic tags to their legs; they then program the array so that only one feeder opens for each bird. Once a chickadee discovers its assigned feeder, the system can count how many incorrect feeders it tries per visit before hitting upon the correct one. The birds that make fewer mistakes are learning faster.

The researchers make regular trips into the mountain range on snowmobiles, to maintain the arrays; they use pulleys to suspend the apparatuses from trees in order to protect the birds from bears and squirrels; they adjust the height as snow accumulates under animals’ paws. They’ve studied about a hundred or two hundred birds a year for the past nine years, and have found that the learning ability of first-year chickadees predicts whether the birds will live through their first winter, which is “a big bottleneck for survival,” Pravosudov said. “Even small differences matter.” In a study that was featured on the cover of Science, this past September, the researchers showed that the worst-performing birds live for about a year on average, while the best performers survive to the age of three.

Memory appears to be shaped not only by natural selection—survival of the fittest—but also by sexual selection. Near the feeder arrays, Pravosudov’s lab maintains more than three hundred and fifty nest boxes, where researchers can monitor chickadee mating and reproduction. In a study published in 2019, they found that, when females mated with males who had aced the memory test, they laid more eggs and raised larger broods than when they mated with those who had fumbled—a sign that the females were investing more in the reproductive encounter. “They’re somehow able to know something about the cognitive abilities of their male,” Carrie Branch, the paper’s primary author, who is now a professor at Western University, in Ontario, told me. Branch thinks that the female birds pick up on a signal—perhaps song complexity, or plumage color—which correlates with cognitive ability. (The paper is titled “Smart Is the New Sexy.”)

Ultimately, food caching relies on spatial memory; this, in turn, depends on genes and the brain. Pravosudov has explored the genetic and brain mechanisms on which food caching depends. Research going back to at least 1966 has shown that damage to the hippocampus—a part of the brain involved in memory—hinders birds from returning to their caches. In the nineteen-eighties, scientists found that food-caching species have larger hippocampi than other species. Pravosudov has extended these findings to incorporate the evolutionary pressure exerted by different environments. In 2002, he and his postdoctoral adviser, the comparative-cognitive scientist Nicola Clayton, compared black-capped chickadees captured in Alaska with birds from Colorado, where winters are slightly less harsh. In the lab, the Alaskan birds outperformed the Coloradans, and their hippocampi were disproportionately larger, and contained more neurons. In later studies, he and a postdoc, Timothy Roth, found similar neural patterns in birds from ten North American locations with climates of varying harshness.

London taxi-drivers, who must memorize complex navigational routes, experience growth in areas of their hippocampi. That’s a matter of training, not genes: their brains have bulked up from use, like biceps. To better understand the significance of nature and nurture, Roth, Pravosudov, and their collaborators hand-raised black-capped chickadees, which they’d plucked from nests in Alaska and Kansas. As before, Alaskan birds outperformed comparison birds on spatial-memory tasks; their hippocampi were the same size but the Alaskan birds had more hippocampal neurons. In 2022, Branch, Pravosudov, and their colleagues managed to identify more than two hundred genes associated with spatial cognitive performance, several of which were known to influence hippocampal development. Both genes and nurture, in short, played a role.

The large-scale goal of all this work has been to demonstrate natural selection in action—a rare feat, given the time scales on which evolution happens. Such a demonstration, Pravosudov told me, requires three interlocking elements: scientists must identify variation in the wild, understand the trait or behavior’s genetic basis, and show that it has consequences for survival or reproduction. Pravosudov believes that his lab is the only one currently capable of completing the jigsaw puzzle for cognitive, rather than physical, traits—not just for birds but for any animal. For this kind of work, he explained, chickadees in the Sierras are close to ideal: “They’re not migratory, they don’t go anywhere.” That allows for special access. “We measure them every year. We know everything about their life. We know about their cognition, who they mate with, how many offspring they produce.” Food-caching birds are “a beautiful model to study,” he said, “because food caching puts such enormous pressure on their cognition.”

“The evolution of cognition by natural selection is a hotly debated question,” Pravosudov pointed out. Some research suggests that performance on tests purporting to measure human general intelligence is roughly fifty per cent genetically heritable. In 2021—two decades after the psychologist Steven Pinker published “The Blank Slate,” which explored the genetic basis of our behavior—Kathryn Paige Harden, a psychologist at the University of Texas at Austin, met criticism when her book, “The Genetic Lottery,” argued that DNA plays a powerful role in shaping socioeconomic outcomes. (“Beneath Harden’s protestations that she’s an egalitarian hides a stealthy affirmation of the old, tenacious view that races and classes are natural kinds,” a biologist and a historian co-wrote, in The New York Review of Books.) Pravosudov tries to avoid such charged debates about I.Q. by sticking to memory. “Memory is simple, right?” he said. “You remember or not. But intelligence, people don’t know what it means. There’s still so many biases.”

Whatever the trait, when studying natural selection, “the bar for evidence is very high,” he went on. That goes especially for “plastic” traits, such as aspects of cognition that can change over the course of a lifetime. “When we say ‘genetic effect,’ people think, Oh, genetics determines precisely your cognition,” Pravosudov said. “No, it doesn’t. But genetics provide the foundation on which you can build.”

Acheetah’s swiftness is thrilling, but it isn’t surprising: their bodies are clearly built for speed. The same goes for a bat’s ability to avoid obstacles in the dark—it’s noteworthy but not surprising, because they have systems for echolocation. The feats of chickadees and other food-caching birds are especially impressive because they are accomplished through cognition, Dmitriy Aronov, a neuroscientist at Columbia University, told me. (Aronov’s lab has discovered “place cells”—neurons that represent specific locations—in the hippocampi of food-caching birds.) “When you look at a chickadee, it’s mostly head, so it makes some sense,” Carrie Branch said. Still, their abilities are startling, in part because we tend to think of memory as a largely human gift.

And yet chickadees may pay a price for their expansive spatial memories. In one study, Pravosudov’s lab tested chickadees that resided at two different elevations—around six thousand and eight thousand feet. Those at the higher elevation faced harsher winters and greater selection pressure to remember where they stashed food. As one would predict, the higher-elevation birds made fewer errors after first learning which feeder in the array opened for them. But then Pravosudov’s team told that feeder to stop responding, so that the birds had to learn to visit a different one, instead. This task, called reversal learning, is a measure of cognitive flexibility: you have to drop what you know and replace it with something else. Now the higher-elevation birds performed worse than the lower-elevation birds. Their memory’s grip was too tight. The higher-elevation birds showed a greater amount of what’s called proactive interference—a phenomenon, well-studied in humans, in which learning one set of information can inhibit the learning of later information. Drilling one tennis swing, for instance, can mess you up when you try to make an adjustment.

Another study, published last year, tested an even trickier problem, learning a rule: on each day, one of two feeders opened for a bird, and the assigned feeders flip-flopped daily. Success meant absorbing the abstract pattern and predicting which feeder would open the next day—a feat that may require not just associative learning but executive functioning, which can include the abilities to assess a situation, make plans, and put them into action. Higher-elevation birds failed to predict where the food would be each day, unlike some of the more forgetful birds from lower elevations. “We think that high birds have such big and strong memories that they draw them back to the previous feeder, no matter what,” Pravosudov said. “Low-elevation birds may not have this, and so that actually makes them smarter.”

Humans could probably benefit from remembering the locations of thousands of objects. But we get more value from cognitive flexibility and executive functioning. Our different brains reflect different evolutionary histories, and allow us to approach survival differently. “If you or I forget our car keys, we’re not going to die of starvation,” Nicola Clayton told me. “We’ve probably got a mobile phone handy, and we can call a loved one or a friend and get things sorted, and get back into the house, and live for many more years thereafter.” We didn’t evolve with mobile phones, of course—but we did evolve in an ecological niche that afforded us multiple ways of solving our problems. The pressures of natural selection rewarded us for navigating between them. “I probably would not be able to remember all of the locations that a chickadee remembers in the forest,” Aronov said. “But on the other hand I remember so many facts, and complex relationships between concepts.” Using these, he can generalize, invent, imagine, and so on.

Still, we might wonder if a person could have a memory a little more like a chickadee’s. In a 1942 short story, “Funes the Memorious,” Jorge Luis Borges imagined a man who, after falling from a horse and hitting his head, obtained “infallible” perception and memory. “He knew by heart the forms of the southern clouds at dawn on the 30th of April, 1882, and could compare them in his memory with the mottled streaks on a book in Spanish binding he had only seen once,” Borges wrote. The man complained, however, of having a memory “like a garbage heap.” He was “almost incapable of ideas of a general, Platonic sort.” It was “difficult for him to comprehend that the generic symbol dog embraces so many unlike individuals of diverse size and form; it bothered him that the dog at three fourteen (seen from the side) should have the same name as the dog at three fifteen (seen from the front).” Borges writes that “his own face in the mirror, his own hands, surprised him every time he saw them.”

A condition called highly superior autobiographical memory (H.S.A.M.) approximates what Borges imagined. In 2006, researchers described a woman named Jill Price, who could recall in detail what had happened to her on any day after her tenth birthday, more or less. Price described having a “split screen” in her head, one playing the present and the other playing the past. “It is non-stop, uncontrollable and totally exhausting,” she told researchers. Happy memories were pleasant, but others led to phobias (of bird droppings, for instance). The researchers reported that Price, although her general intelligence was not below the average range, had “significant deficits in executive functions involving abstraction, self-generated organization and mental control.”

Human memory, it seems, was never meant to be perfect and permanent. In a book titled “Forgetting: The Benefits of Not Remembering,” the Columbia neurologist Scott A. Small suggests that the behavioral inflexibility seen in autism may result from genes that diminish forgetting. In a review paper, “The Many Faces of Forgetting: Toward a Constructive View of Forgetting in Everyday Life,” the psychologists Jonathan Fawcett and Justin Hulbert list other documented benefits of forgetting: creativity, self-reinvention, mental clarity, big-idea thinking, forgiveness, the updating of knowledge, and the reduction of post-traumatic stress. During sleep, they note, the brain prunes its neural connections. (Machine-learning researchers use similar tactics to push artificial neural networks away from the data on which they’ve trained and toward more useful generalizations.) “Some researchers have gone so far as to claim that forgetting is the ‘default mode’ of the brain,” Fawcett and Hulbert write.

And yet there may be commonalities between bird and human memory. In searching for my beard trimmer, I used so-called episodic memory—a kind of recollection comprising replayable experiences, as opposed to facts, associations, or skills. Episodic memories have a what, a where, and a when. (The last time I saw my beard trimmer, I was . . .) Researchers debate the question of whether birds use episodic memories: perhaps, instead of drawing on a simple list of locations, they see a tree and recall the time they stored a seed there. In the nineteen-nineties, Clayton conducted pioneering work that succeeded in demonstrating something like episodic memory in scrub jays. In a lab study, birds prioritized the retrieval of freshly buried worms over peanuts; then, after five days, they switched to seeking peanuts. They seemed to know that certain locations contained worms that they had buried five days earlier, which would be rotten. (They weren’t relying on smell, as the food had been removed.) In other words, they remembered not only where they had cached food but also what they had buried, and when. Clayton’s paper, published in Nature, described the finding as “the first conclusive behavioural evidence of episodic-like memory in animals other than humans.”

Some researchers believe that episodic memory entails consciousness. (How can a scene be reconstructed in memory if there’s no one there to rewind and watch it?) Clayton and Pravosudov told me that they lean toward believing that birds have conscious awareness (although, ultimately, they leave the question to philosophers). Even if birds are conscious, though, I wouldn’t want to remember like a bird. I prefer the special adaptations of human thinking—adaptations that allow for language learning, empathy, and the ability to adjust and abstract. Survival for us goes far beyond collecting, hiding, and retrieving seeds. We must use tools, coöperate, change our beliefs. A mind like a steel trap would be nice—but I prefer one like a shopping cart.