Bumblebees can learn to solve puzzles from experienced peers. Honeybees do the same to learn their waggle dances.
Enlarge / Bumblebees can learn to solve puzzles from experienced peers. Honeybees do the same to learn their waggle dances.

Social insects like bees demonstrate a remarkable range of behaviors, from working together to build structurally complex nests (complete with built-in climate control) to the pragmatic division of labor within their communities. Biologists have traditionally viewed these behaviors as pre-programmed responses that evolved over generations in response to external factors. But two papers last week reported results indicating that social learning might also play a role.

The first, published in the journal PLoS Biology, demonstrated that bumblebees could learn to solve simple puzzles by watching more experienced peers. The second, published in the journal Science, reported evidence for similar social learning in how honeybees learn to perform their trademark “waggle dance” to tell other bees in their colony where to find food or other resources. Taken together, both studies add to a growing body of evidence of a kind of “culture” among social insects like bees.

“Culture can be broadly defined as behaviors that are acquired through social learning and are maintained in a population over time, and essentially serves as a ‘second form of inheritance,’ but most studies have been conducted on species with relatively large brains: primates, cetaceans, and passerine birds,” said co-author Alice Bridges, a graduate student at Queen Mary University of London who works in the lab of co-author Lars Chittka. “I wanted to study bumblebees in particular because they are perfect models for social learning experiments. They have previously been shown to be able to learn really complex, novel, non-natural behaviors such as string-pulling both individually and socially.”

Ever since Charles Darwin observed bumblebees in 1884 engaged in “nectar robbing”—in which a forager bites into the base of a flower to collect the nectar but doesn’t pollinate the plant—scientists have recognized a surprising capacity for social learning in creatures of all kinds. An evolutionary theory called the Baldwin effect says that certain beneficial behavioral traits learned in the lifetime of one creature are passed on to its offspring via natural selection. So Bridges et al. decided to explore the possibility that social learning might have contributed to unique behavioral innovations in bumblebees (Bombus terrestris), rather than being purely instinctive.

As we reported previously, Chittka’s group conducted a 2017 study in which it showed that bees could be trained to roll little wooden balls in order to receive a reward. But the group also noticed instances where the bees opted to roll the balls even when there wasn’t an obvious reward or benefit. The balls had been placed in a tunnel that connected the hive to the experimental arena where the food was. Several bees walked over the balls or stopped to roll them on their way back and forth from the food. The group wondered if this might be genuine play behavior.

Chittka’s lab published a follow-up study last year reporting its observations of genuine play behavior in bees, which were filmed rolling small colored wooden balls. (While many animals are known to engage in play, they are usually large-brained mammals and birds.) For Chittka, this was “a strong indication that insect minds are far more sophisticated than we might imagine.”

The idea to use two-option puzzle box experiments to explore social learning came from prior research on chimpanzees (2005) and great tits (2015). For the bees, the Chittka lab designed puzzle boxes that could be opened by rotating a clear lid, either by pushing clockwise on a red tab or counter-clockwise on a blue tab. This would let the bees collect a tasty reward of 50 percent sucrose solution.

A bee opens a puzzle box by pushing against the red tab to rotate the lid of the box clockwise. Credit: A.D. Bridgs et al., 2023/CC-BY 4.0

A demonstrator bee was trained to perform one of the two solution behaviors and then added to a group of untrained bees. All bees were allowed to forage freely among the puzzle boxes, and their behavior was monitored to see if the demonstrator bee repeatedly “solving” the puzzle box resulted in that behavior spreading through the rest of the group. There were also control groups with no demonstrator bees.

“We wanted to see whether bumblebees could learn a non-natural, novel behavior via social learning under open diffusion conditions,” said Bridges. “In particular, we were interested in whether they could learn variations of this behavior that were specific to their trained demonstrator—just like these larger-brained species.”

That’s exactly what happened. Foraging bees in colonies that had demonstrator bees opened far more puzzle boxes than the control bees, and they used the solution they had been “taught” 98 percent of the time, suggesting they had learned the behavior socially. In follow-up experiments with multiple demonstrator bees, each teaching a different solution to the puzzles, the untrained bees initially learned both methods but, over time, preferences for one or the other randomly developed, much like a popular fad or trend. Once there was a preferred solution, the bees usually stuck with it.

As for the two control colonies, only one bee learned to open the puzzle boxes—a “spontaneous learner,” if you will—although it was less proficient at doing so and opened far fewer boxes than bees in colonies with a demonstrator. A second round of experiments that ran longer produced a few more spontaneous learners, also less proficient. That said, one spontaneous learner bee was an overachiever, with 216 recorded box openings over the course of the experiment; its closest rival in the control groups only opened 22 boxes over the same time period. By contrast, the trained demonstrator bees were expert foragers, routinely opening over 100 puzzle boxes every day.

The findings are consistent with the prior studies involving chimpanzees and great tits, in which the authors argued that such social learning demonstrated those species’ capacity for culture. “Bumblebees aren’t known to show culture-like phenomena in the wild,” said Bridges. “Even so, our bees appeared to have been capable of this. I was also surprised by how flexible and innovative bumblebee behaviour really is. The fact that some bees learned to open the boxes by themselves was really remarkable and definitely suggests that bees might come up with other new behaviors in the wild—as long as the opportunity and need arises.”

Further studies on the nectar-robbing behavior of bumblebees could shed more light on the underlying mechanics. But can this socially learned behavior be passed down to subsequent generations? That’s harder to test with this particular type of bumblebee. “The bumblebees in our experiments form colonies that last for a single biological generation before collapsing, but honeybees, stingless bees, and some tropical bumblebees form colonies that last for many years,” said Bridges. “If naturally occurring, long-lasting culture were to be found in an invertebrate, it would probably be here.”

Waggle waggle

Honeybees perform a waggle dance. Credit: Shihao Dong et al., 2023

That’s where the second study on waggle dances in honeybees comes in. “We are beginning to understand that, like us, animals can pass down information important for their survival through communities and families,” said James Nieh of the University of California San Diego, a co-author of this study. “Our new research shows that we can now extend such social learning to include insects. We know that bees are quite intelligent and have the capacity to do remarkable things.”

Honeybees use the waggle dance to tell other bees where to find critical resources like food, water, tree resins, and nest sites. The “dance” involves waggling their abdomens in the shape of a figure eight, and it’s done at a remarkably fast pace: A bee can move an entire body length in less than one second. Longer waggle runs indicate longer distances, while the angle of the waggle’s direction indicates direction. The quality of the resource in question is encoded in the number of waggle run repetitions and in how quickly each dancing bee comes back to repeat the successive runs.

Per Nieh et al., there is a strong genetic component to the waggle dance behavior in honeybees, but the team hypothesized that novice dancers could improve their skill at communicating with their fellow bees by learning from more experienced waggle dancers. Workers start following waggle dancing foragers at around eight days old and start dancing themselves at 12 days, usually dancing the same “routines” as the older bees they shadowed. Nieh et al. figured that those first early waggle dances would thus have more communication errors if the bees were raised without being able to follow older dancers. So they reared colonies of honeybees with young bees all the same age and monitored the emergence of waggle dance behavior, along with mixed control colonies with older experienced bee dancers.

The results supported their hypothesis. The first waggle dances of bees who weren’t able to shadow older experienced forager bees were riddled with errors regarding the direction and distance of valuable resources. While they were able to improve through practice or watching other bees, they were never able to correctly encode distance. Bees in the mixed control colonies gained valuable social cues by following experienced dancers, so their first dances were far more accurate. And that social knowledge stayed with them for their entire lives.

In essence, the untrained bees developed their own unique distance dialect. “Scientists believe that bee dialects are shaped by their local environments,” said Nieh. “If so, it makes sense for a colony to pass on a dialect that is well adapted to this environment.” He and his co-authors believe their results suggest that social learning shapes honeybee signaling, much as it does in many vertebrate species.

“It is at least plausible that there might be subtly different, socially acquired local ‘cultures’ of the dance language that depend on visual characteristics of the landscape or the spatial distribution of food sources,” Chittka and Natasha Rossi (of the University of Sussex in Brighton) wrote in an accompanying commentary. They cited a 2008 study showing that one species of honeybee learned to read through the distance code of another species, as well as Chittka et al.’s own study on playful ball rolling in bees. “It is therefore plausible that some of their most advanced behavioral innovations (including elements of the dance language) might have emerged at least in part by individual innovation and subsequent social learning, becoming instinctual later in evolutionary time.”

DOI: Science, 2023. 10.1126/science.ade1702  (About DOIs).

DOI: PLoS Biology, 2023. 10.1371/journal.pbio.3002019  (About DOIs).