The waggle dance of the western honeybee (Apis mellifera), in which bees waggle their abdomen from side to side while repeatedly walking in an intricate figure-of-eight pattern, has been observed since antiquity, but the person who finally unlocked the secret of its meaning was an iconoclastic Austrian researcher named Karl von Frisch. The breakthrough initially earned Frisch a great deal of scorn from other mid-20th-century scientists, but also eventually won him the Nobel Prize.
Young Karl was known to skip school to spend time with a menagerie of over 100 animals, only nine of which were mammals. His most beloved companion was a small Brazilian parakeet named Tschocki, who was constantly by Frisch’s side, sitting on his lap or on his shoulder and even sleeping next to his bed. Together with Tschocki, Frisch spent hours out in nature, simply watching. As he later reflected: “I discovered that miraculous worlds may reveal themselves to a patient observer where the casual passerby sees nothing at all.”
Frisch began studying bees in 1912. He had a hunch that ran counter to prevailing wisdom: The bees’ waggle dance was a form of language. In pursuing this hypothesis, he was contesting two core assumptions of Western science and philosophy: that only humans have complex forms of language, and that insects were incapable of complex communication given their tiny brains.
Human verbal language is largely based on the noises we make with our vocal cords and mouths, the expressions we make with our faces, and the way we hold and move our bodies. In contrast, bee language is mostly spatial and vibrational. Its syntax is based on something very different from human language: the type, frequency, angle and amplitude of vibrations made by the bees’ bodies, including their abdomens and wings, as they move through space.
By buzzing and quivering, leaning and turning, bees communicate remarkably accurate information. Once a scout bee has found a good food source, she returns to the hive to inform her sisters. During the waggle dance, the bee moves in a figure eight pattern: a straight line while beating wings, and then a circular return without wing beating. We know now that the resulting pattern, which can be observed visually, encodes the direction to the food source relative to the sun’s position in the sky; the length of the dance is related to the distance the bees must travel.
“Miraculous worlds may reveal themselves to a patient observer where the casual passerby sees nothing at all.”
Frisch decided on an ambitious experimental design: tracking thousands of individual bees in order to analyze the correlation between their dances and specific food sources. At the time, this seemed impossible, given that hive populations average somewhere between 10,000 and 40,000 bees. But Frisch, through painstaking attention to detail and near endless amounts of patience, was able to prove his hypothesis: As a lead bee dancer waggles, she orients her body relative to gravity and the position of the sun. By making subtle variations in the length, speed and intensity of her dance, she is able to give precise instructions about the direction, distance and quality of the nectar source. In so doing, she teaches other bees in the hive, who use the information they have learned from the waggle dance to fly to a nectar source they have never before visited.
Frisch’s research progressively proved the astonishing accuracy of the bees’ communication system. In one of his most famous experiments, he trained his bees to navigate to a hidden food source several miles away, across a lake and around a mountain. This was an astounding feat, given that he had shown the site once to only a single bee. In another experiment, he demonstrated that different hives have slightly different dancing patterns. Bees appeared to learn these patterns from their hive mates. In essence, honeybee dance language has dialects, just like human communities.
Frisch himself was so amazed by his findings that he initially kept them secret. Contradicting prevailing scientific views, his findings demonstrated that honeybees possessed learning, memory and the ability to share information through symbolic communication, a form of abstract language. As he wrote to a confidante in 1946: “If you now think I’m crazy, you’d be wrong. But I could certainly understand it.”
Frisch was right to worry. When he finally went public, many scientists dismissed his research and argued that insects with such tiny brains were incapable of complex communication. The American biologist Adrian Wenner launched a challenge to Frisch’s theory, arguing that bees locate foods solely by odors, a theory that was subsequently proved wrong, although odors are important signals for bees. Eventually, Frisch’s results were definitively and independently validated, and he was awarded the Nobel Prize in 1973. The prize committee concluded its nomination statement by referring to the “shameless vanity” of Homo sapiens that refused to recognize bees’ extraordinary capacities.
Frisch referred to honeybee dances as a “magic well”: The more he studied them, the more complex they turned out to be. Every species, Frisch argued, has its own magic well. Humans have verbal language. Whales have echolocation, which endows them with the ability to visualize their entire environment via sound. Honeybees have spatial, embodied language: We now recognize some of the subtle differences in their body movements and vibrations, which include waggling, knocking, stridulating, stroking, jerking, grasping, piping, trembling and antennation, to name just a few.
The bees’ dance is still considered by many scientists to be the most complex symbolic system that humans have decoded to date in the animal world. Although many scientists initially asserted that the waggle dance should be referred to merely as communication, Frisch insisted on using the term language: Through a system of signs, bees exchange information, coordinate complex behavior and form social groupings.
Honeybee researchers following in Frisch’s footsteps have probed the magic well even more deeply. Bees make many other types of signals through nuanced movements, communicating through sounds and vibrations largely either inaudible to or indecipherable by humans. Moreover, by using computer software that automates the decoding of bee vibrations and sounds — vibroacoustics, as the field is known — researchers are now using algorithms to analyze bee signals. Their discoveries are as incredible as Frisch’s first breakthroughs.
Although it has been known for centuries that queens have their own vocabulary (including tooting and quacking sounds), researchers have found new worker bee signals, such as a hush (or stop) signal that can be tuned to specific types of threats and a whooping danger signal that can be elicited by a gentle knocking of the hive. Worker bees also make piping, begging and shaking signals that direct collective and individual behavior.
“Honeybees exhibit sophisticated forms of democratic decision-making.”
Bees have excellent eyesight and are capable (after minimal training) of distinguishing between Monet and Picasso paintings. They can differentiate not only between flowers and landscapes but even human faces, demonstrating a remarkable capacity for processing complex visual information. In two breakthrough experiments in 2016 and 2017, researchers demonstrated that bees are capable of social learning and cultural transmission (a first in Western science for invertebrates): When trained to pull a string to receive a sugar reward (a novel task), bees taught the new skill to their hive mates, demonstrating that bees can learn from observing other bees, and that these learned skills can be shared and become part of the culture of the colony.
A dark side of bee social life has also been uncovered: While honeybees are generally collaborative, accurate and efficient, they are also capable of error, robbery, cheating and social parasitism. They might even have emotions, exhibiting both pessimism and dopamine-induced mood swings that are analogous to human highs and lows.
As one researcher cautiously noted in a landmark study of a newly identified bee signal: “Communication in honeybees turns out to be vastly more sophisticated than originally imagined. Research is revealing … a collective intelligence that … makes one pause to ask whether these creatures may be more than just simple, reflexive, unthinking automata.”
Perhaps the most remarkable research is that of Cornell bee scientist Thomas Seeley, who has demonstrated that honeybee language extends beyond foraging behavior. For several decades, Seeley focused his research on bee swarming. Swarming is the way honeybee colonies naturally reproduce; a single colony splits into two or more distinct colonies, and one group flies off to find a new home. How, Seeley wondered, did the colony decide on their preferred site?
When Seeley first decided to focus on swarming, scientists knew very little about the phenomenon. The fastest bees in a swarm fly over 20 miles per hour, usually moving in a straight line toward their target regardless of the fields, water bodies, buildings, hills or forests in their way. There is no way a human can keep up with the swarm, much less keep track of several thousand individual bees to figure out which ones, if any, are guiding the rest. Seeley was interested in how the bees decided which home to select — a high-stakes decision, given that splitting the hive could cause the queen to be lost, and choosing an inappropriate site could lead to the death of the hive.
In the mid-2000s, Seeley convinced a computer engineer who was intrigued by the similarities between bee swarms and driverless cars to install a high-powered video camera at Seeley’s research site on Appledore Island, off the coast of Maine. Their goal was to create an algorithm that could automatically identify and track some 10,000 speeding bees at once.
“Communication with bees is an ancient human skill.”
After two painstaking years, the algorithm finally worked: Powered by high-speed digital cameras and novel techniques in computer vision, it could identify each individual bee from the video footage and analyze its unique frenzied flight pattern. The algorithm revealed patterns undetectable to the human eye; decoding the diversity, density and interactions in these patterns led Seeley to label the swarm as a “cognitive entity.”
Perhaps Seeley’s most startling finding was that, in choosing a new home, honeybees exhibit sophisticated forms of democratic decision-making, including collective fact-finding, vigorous debate, consensus building, quorum and a complex stop signal enabling cross-inhibition, which prevents an impasse being reached. A bee swarm, in other words, is a remarkably effective democratic decision-making body in motion, which bears resemblance to some processes in the human brain and human society. Seeley went so far as to claim that the collective interactions of individual bees were strikingly similar to the interactions between our individual neurons when collectively arriving at a decision.
Seeley’s findings bolstered the arguments of those who argued in favor of referring to honeybee communication as language. And by demonstrating that the “hive mind” was more than mere metaphor, Seeley also stimulated advances in swarm intelligence in robotics and engineering. Seeley’s research, predicated on digital technology (computer vision and machine learning) eventually came full circle: His findings inspired computer scientists at Georgia Tech to create the Honey Bee algorithm, which is now an integral part of cloud computing: In internet hosting centers (analogous to hives), it optimizes the allocation of servers (foraging bees) among jobs (nectar sources), thereby helping to deal with sudden spikes in demand and preventing long queues. In 2016, Seeley and his collaborators were awarded the Golden Goose Award, which recognizes apparently esoteric research that later proves to be extremely valuable.
Dancing Honeybee Robots
Thanks to Frisch and his successors, researchers have long known that bees react differently to distinct vibration patterns that act like signals. In the past few years, the combination of computer vision with miniaturized accelerometers (ultrasensitive versions of the motion-detecting sensors in your cell phone) has enabled scientists to decode the specific subtle vibration signals made by living organisms — vibrations that are vital to their communication but largely undetected by humans. Indeed, these technological advances have made it possible to analyze bees’ communication and activity over their entire lifespan.
The next breakthrough — bridging what engineers call the “reality gap” between robots and living bees — is the creation of robots that accurately mimic these vibration patterns. Tim Landgraf, a professor of mathematics and computer science in Berlin, has devoted himself to this task for the past decade. Much of his research has focused on automating identification of individual bees and tracking their movements using computer vision and machine learning. One experiment analyzed around three million images taken over three days and tracked the trajectories of every single member of a honeybee hive — with only a 2% error rate.
Landgraf’s most innovative work involves creating robotic devices to communicate with honeybees in their own language. Working together with colleagues in the Free University of Berlin’s Center for Machine Learning and Robotics, Landgraf built a simple robot, which they christened RoboBee. Early prototypes “sucked,” as Landgraf put it: The bees would attack them, biting, stinging and dragging them out of the hive.
“Honeybee dance language has dialects, just like human communities.”
The seventh prototype was the breakthrough. A statistically significant number of bees would follow the RoboBee’s dance and then fly to the specific location that Landgraf had coded into his honeybee robots. He had created, in essence, a bio-digital equivalent to Google Translate for bees.
Some of his robots’ commands are successful with the bees and sometimes not, and Landgraf still isn’t sure why. His current hypothesis is that a separate, prior signal needs to be issued first, like a handshake before a conversation can begin. His robotic bees may sometimes be emitting this signal merely by chance, and in those cases the bees in the hive will listen. Or perhaps a separate vibrational signal from a different device is also needed; one such tool, recently invented by Cornell bee researcher Phoebe Koenig, accurately mimics the “shaking” signal that bees use to activate behavior.
One day, he hopes, the RoboBees will be viewed as “native” by the honeybees themselves, able to issue commands and recruit bees to fly to specific locations by waggle dancing. Future robots might even learn local bee dialects, which vary with habitat. And this is only the tip of the iceberg; his work could enable the possibility of understanding how the colony itself processes and integrates different kinds of information, somewhat like a living distributed computer with thousands of tiny, interconnected brains.
Landgraf is now going beyond bee monitoring and trying to build smart hives that are two-way communication devices. Vibrational, acoustic and pheromone signals could be released to warn bee colonies about threats (such as nearby fields treated with pesticides, or approaching storms) or to guide bees to find the best food sources available.
Honey Hunters
As groundbreaking as these innovations might sound, Landgraf is not the first to have discovered how to speak to bees using vibroacoustics. Communication with bees is, in fact, an ancient human skill.
The earliest known vibroacoustic device, the bull-roarer, is regarded as one of humanity’s oldest musical instruments. Used in ceremonies by Indigenous peoples on all continents and in the Dionysian Mysteries by the ancient Greeks, it has a lesser-known function as a bee-hunting device. A bullroarer (turndun or bribbun to Australian Indigenous communities, kalimatoto padōk to the Pomo tribe in California) is deceptively simple: A long string or sinew is attached to a thin, rectangular piece of wood, stone or bone that is rounded at the ends. The cord is given a slight initial twist, and then the bullroarer is swung around in a circle. The resulting noise, caused by air vibrating between 90 and 150 Hz, is a surprisingly loud sound similar to a propeller. The effect is startling and palpable: a resonating hum in your bones, like standing within a giant swarm of bees.
Africa’s /Xam (San) use bullroarers to cause bees to swarm and to direct them to new hives at locations that are easy for humans to access. The /Xam word for bullroarer is “!goin !goin,” which literally means “to beat” — like beating a drum. The bullroarer is spun in tandem with a dance that puts the /Xam into a trance-like state through which elders call upon and guide the bees. (Modern beekeeper practice employs a simple version of this method, called tanging, to calm bees and direct them to a hive.) Long before Western science discovered vibroacoustics, the /Xam had developed a nuanced understanding of bee communication. Anthropologists speak of a “copresence” that the /Xam developed with bees, based on mimetic sound capacities.
The /Xam are not unique in their ability to communicate with bees. In parts of Africa, people searching for honey are led to beehives by a bird: the greater honeyguide (its Latin name, Indicator indicator, is a bit of a giveaway). Honey hunting is an ancient art; some of the earliest recorded rock paintings in the world show humans hunting wild bees. And the animal kingdom’s preeminent honey hunters are honeyguides.
“Honeybees can differentiate not only between flowers and landscapes but even human faces, demonstrating a remarkable capacity for processing complex visual information.”
Honeyguides are one of the only birds (and few vertebrates) on the planet that eat beeswax. Rich in nutrients and energy-giving lipids, beeswax is a sought-after treat for the birds. But most honeybee nests in Africa are well hidden in tree cavities, guarded by fierce bees that can kill the birds if they come too close. Honeyguides — likely guided by their strong sense of smell — know where the bees are but can’t get at the wax. So they partner up with an animal that isn’t nearly as good at finding bees but knows how to get the wax: humans.
In hunting together, the honeyguides and honey hunters have evolved a subtle form of cooperative communication. First, the hunters make their special call, signaling that they are ready to hunt honey. In the case of the Yao hunters in the Niassa National Reserve in Mozambique, who were the focus of researchers led by Claire Spottiswoode at Cambridge University, this sound is something like a brrr-hmmm: a loud trill followed by a grunt. In return, the honeyguides approach and sing back to the hunters with a special chattering call.
The birds then fly in the direction of the bees’ nest, followed by the hunters. When the birds’ chatter dwindles and they stop flying, the hunters know they are close. They scan the tree branches and hit nearby tree trunks with their axes to provoke bees into revealing the location of the nest. The hunters then make a bundle of leaves and wood and set it alight just under the nest, smoking the bees into lethargy before felling the trees with their axes and chopping open the nest. As they fill buckets to take back home, flinging away dry combs containing no honey, they expose food for the birds. The honeyguides wait patiently, flying down to feed only after the humans are gone. Before the Yao hunters depart, they gather up the wax and present it on a little bed of fresh green leaves, honoring the contribution of the birds to their hunt.
Scientists have confirmed the claims of the Boran people in northern Kenya that they can infer the distance, direction and time to the nest from the bird’s calls, perching height and flight patterns. Spottiswoode also confirmed reciprocal signaling among the Yao: When honey hunters made their special sound, the probability of being guided by a honeyguide increased from 33% to 66%, and the overall probability of finding a bees’ nest from 17% to 54%.
We might expect the ability to interpret human sounds from trained animals like falcons and dogs, and even some wild animals like dolphins, but from wild birds? The sounds exchanged between hunters and honeyguides are also not the same across Africa. They are learned from elders, passed down from one generation to the next. How do birds learn to communicate with humans? We don’t actually understand this yet, but we do know that honeyguides don’t learn cooperative hunting from their parents. Honeyguide nestlings never meet their parents, as the species is brood parasitic: Adults lay their eggs in other birds’ nests, puncturing any host eggs they find to enhance the honeyguide hatchlings’ survival rate. Then the adult honeyguides leave. Right from birth, honeyguide hatchlings are equipped with sharp, hooked beaks, which they often use to kill any unfortunate host chicks that managed to survive.
So how do the honeyguides learn the sounds? Spottiswoode and her colleagues are combining digital technologies with traditional knowledge to find out. They have developed a customized app that enables honey hunters to collect data on their activities. Deep in the forests of the Niassa, an area the size of Denmark with few roads and no internet connectivity, Yao honey hunters are roaming the forest armed with handheld Android devices, earning income from Cambridge University as digital conservation research assistants, singing to their honeyguide companions as they search for bees.
Governing The Swarm
Proponents of smart hives argue that digital technologies offer the potential to enhance environmental protection in a partnership between humans, insects and AI-enabled robots. Smart hives could use sensors and cameras to monitor bees and provide them with information to guide crop pollination and avoid polluted sites. The same technologies might be used to harness bees to map zones too dangerous for humans to reach, or power swarm robots to support environmental conservation, or even help out with search-and-rescue missions.
As data accumulates, a twinning effect emerges, with some beehives now also existing virtually in digital bee world that mirrors the physical one. This may help turn the tide in our race to save not only honeybees but many other species as well. When gathering nectar, bees continuously sample from the environment, so who better to act as a sentinel for environmental risk? Bees and other insects have been successfully trained to detect a range of chemicals and pollutants. Decoding a large number of dances from a specific area could help evaluate landscapes for sustainability and conservation. It could also make pollination more efficient and provide insights into how to ward off the widespread, alarming phenomenon of colony collapse disorder. Bees could also be recruited as live bioindicators: surveying, monitoring and reporting the landscape in a fine-grained, inexpensive way that would be impossible for humans to achieve alone.
But these technologies also create opportunities to weaponize bees. Bees have a long history with the military, and recently they have become instrumental in some security objectives. In the United States, the military has been actively testing bee bio-detectors in antinarcotics, homeland security and demining operations. The mobilization of what military scientists call “six-legged soldiers” requires genetic and mechanical manipulation of the bees’ nervous systems, migration patterns and social relationships.
The Stealthy Insect Sensor Project, for example, trains bees to extend their tongues when they detect dangerous chemicals. As Jake Kosek writes in Cultural Anthropology, once trained, individual bees can be used in military monitoring devices. Trained bees are inserted into cartridges in monitors carried by soldiers. When bees react to, say, military-grade explosives, the microchip in the monitor translates this signal into an alarm. The trained bees live for no more than a few weeks, dying within the cartridge. A replacement cartridge is shipped to the soldier, and, according to the scientist responsible for the project, “you simply slip out one bee cartridge and replace it with another.”
“Technology-driven advances in our ability to understand how bees live and communicate should hardly be repurposed to turn them into living tools for conducting warfare.”
Mobilizing bees to detect dangerous explosives might be beneficial for military personnel, but the manipulation and casual disposal of honeybees at scale should give us pause. Technology-driven advances in our ability to understand how bees live and communicate should hardly be repurposed to turn them into living tools for conducting warfare.
There are other ways of thinking about our relationships with bees. For traditional cultures like the /Xam and the Yao, communicating with bees is embedded in sacred ceremony. Honey is both a practical and a spiritual matter, both food and sacrament. This view is not limited to hunter-gatherers in Africa; the earliest Neolithic representations of bee goddesses from Europe are over 8,000 years old. And many of humanity’s oldest written texts celebrate bees’ divinity. Almost 3,000 years ago, the scribes of the Brihadaranyaka Upanishad, a key text in Hinduism, recorded the “Honey Doctrine” — a theory of the organic, interrelated nature of life, wherein honey personifies cosmic nourishment for the luminous ground of being: “this earth is honey for all creatures, and all creatures are honey for this earth.”
To witness biohybrid bees engaging in reciprocal (if rudimentary) interspecies communication gives me a numinous sense of awe. To witness bees being converted into disposable, militarized sensing devices gives me a sense of dread. These two choices are emblematic of humanity’s relationship with nature. Will we choose dominion or kinship?
If we choose the latter, there is likely to be a great deal more for bees to say to us and for us to say to them. And they will not be the only species that humans engage in dialogue.
