Given a flower just out of reach and a ball they had never been taught to use as a step, bumblebees put two and two together. The experiment suggests that even a sesame-seed-sized brain can combine familiar pieces into a new solution.
A bumblebee stands beneath a flower it cannot reach. Flying is impossible: the ceiling is too low. Stretching gets it nowhere. But a small white ball lies nearby.
The bee begins to roll it across the floor, parks it beneath the flower and climbs on top. No one has taught it this particular trick.
A new study published in Science has cognitive researchers buzzing – the humble backyard bumblebee can draw on previous experience to solve a surprisingly complex new puzzle. The bees already knew that the blue artificial flower offered a sugary reward and that the ball could be moved. But they had never been trained to use it as a stepping stool – and some even rolled it towards a destination they could no longer see.
“Most people still think that insects do not learn anything – that they are almost like robots,” says co-author Olli Loukola, a behavioural ecologist who leads the Bee Behaviour and Cognition Group at the University of Oulu and is also a Senior Researcher at the University of Turku.
“With this study, we actually show that insects are worthy of respect, even though they have brains the size of sesame seeds,” says co-author Ece Nur Akmeşe, who logged hundreds of hours observing bees for the study an undergraduate at the University of TOBB Economics and Technology in Türkiye
Sisyphus who?
Previous studies by Loukola and his team had demonstrated that bumblebees can learn to scoot around a sphere, even though bees have no known ecological or evolutionary history of rolling detached objects to reach flowers. “This is not natural for bees,” Loukola says. In fact, the harder part can be getting them to stop. The bees remain enthusiastic even when there is no reward, offering researchers a peek at play and what insects think is fun.
This enthusiasm also complicates the detective work: a ball reaching the right place could reveal a bee solving a problem – or merely doing something it already enjoys.
In 2016, researchers at Queen Mary University of London in the United Kingdom also found that bees could be taught to pull a string to get a tasty sip of sugar water. Together, the studies showed that bees can learn unusual actions and sometimes perform them for their own sake. But could they cobble familiar pieces together into a solution no one had shown them before?
The team devised a series of tests to determine whether the bees could put two and two together – or whether they were merely following the flower with their eyes, learning by trial and error or rolling into the right place by chance.
The bees knew the cast – but not the plot
First, they taught one group of bumblebees that sugary treats were present on blue circles – a sort of simplified flower – and let them become familiar with movable extruded polystyrene foam balls. At one stage, the bees had to nudge a ball off the flower to reach the sugar beneath it.
Other bees learned only that the flower held sugar, and a control group had met neither flower nor ball. So the fully prepared bees knew the cast. What none of them had ever seen was the plot: move the ball beneath a raised flower, climb aboard and collect the reward.
Then, the blue circle was moved just out of reach, onto a ceiling that was too high for the bees to stretch towards but “annoyingly low” enough to leave no room to go airborne, says lead author Akshaye Bhambore, a doctoral researcher in behavioural ecology at the University of Oulu.
About three quarters of the fully prepared bees figured out how to get a boost, steering the ball into a little divot beneath the blue circle and climbing aboard as though it were a stepladder. Bees that knew only that the flower contained sugar – or had never met either piece – were far less likely to succeed.
“Knowing that the flower meant food was not enough. The bees also needed experience with the movable ball. They had all the pieces, but no one had shown them how to put them together to solve the problem,” says Bhambore.
The team was not yet ready to declare victory. This first test showed that the bees could solve the puzzle but not what guided them. Because they could still see the flower while rolling, they might simply have been steering towards a visible target rather than acting towards a remembered destination.
Putting doors between the bees and the answer
So the team began putting doors between the bees and the answer. Behind one opaque barrier, 16 of 22 bees still piloted the ball through an opening and beneath the flower. Nevertheless, the goal may have slipped back into view as the ball approached the doorway, leaving a crack in the case: sight could still have guided the final stretch.
Next came a veritable maze of three chambers, designed to keep the flower hidden throughout the journey. It looked like the toughest puzzle – but proved scientifically inconclusive. “Bees rolled balls into the final chamber at similar rates whether or not a flower was waiting there. Ball-rolling enthusiasm, it seemed, could imitate a plan remarkably well.
The cleanest test gave the bees two concealed compartments.
“They first saw which side held the flower and then had to move the ball towards that remembered location after the flower was hidden,” says Bhambore.
The flower was concealed behind identical walls, and the trial ended as soon as the bee rolled into either compartment – leaving no chance to try the wrong side and then correct course. Twenty-three of 30 chose the compartment associated with the flower, including 16 that headed there directly.
Across the tests, one easy explanation after another had either been ruled out or exposed: following a visible flower, learning through repeated attempts or simply rolling for the fun of it. The final test provided the strongest evidence that the bees were directing the ball towards a remembered location.
To the researchers’ further surprise, a handful of bees solved the problem without using the ball at all. Some managed to parkour their way onto the ceiling, “like Spider-Pig from the Simpsons,” Bhambore says with a grin. The team suspects that the resourceful bees wedged their sharp toenails – properly called tarsal claws – into tiny cracks in the Plexiglas and levered themselves up.
“You should have seen our group text,” Akmeşe says.
“Technically, they failed the task because they did not use the ball,” Bhambore adds. They could not count as successes in the intended ball-rolling task – but they still deserved full marks for ingenuity.
The secret life (of the mind) of bees
What do these findings add to our understanding of the secret internal life of bumblebees? The results suggest that they can take two things learned separately – a rewarding flower and a movable ball – and combine them into a new sequence of actions.
They could even continue towards the spot where the flower had been after it disappeared from view. This is harder to square with the old picture of insects as rigid stimulus-and-response machines.
Loukola says the results show that “they have the flexibility to cope in changing environments”. Such mental versatility could prove useful for bees facing habitat loss and an onslaught of pesticides suspected of disrupting their foraging, mating and cognitive performance. But the laboratory puzzle did not test whether this flexibility actually helps bees survive these threats.
The results also raise the possibility that the bees have a rudimentary form of body awareness – “they realise, ‘I am too short, I cannot reach this flower on my own,’” Loukola explains.
But this remains a tantalising possibility rather than a conclusion. The study was not designed to prove that bees understand the dimensions of their own bodies, and more targeted experiments would be needed before comparing their abilities directly with the spatial self-understanding that young children develop.
What does aha! look like on a bee?
The scientists also stopped short of claiming that they had caught bees experiencing true insight – the sudden flash in which an animal appears to see a problem differently and produces a new solution without prolonged trial and error.
The evidence converges on goal-directed behaviour and not on a particular inner experience. The bees behaved as though they were working towards a destination held in mind, but the team could not see whether the mental gears shifted suddenly, gradually or at all.
To clear this higher hurdle, they would need to catch the transition itself: a bee apparently stuck before the solution, then behaving recognisably differently after the answer clicks.
“We are trying to document that ‘aha!’ moment,” Loukola says. But since bees do not shout ‘Eureka!’, the team will have to turn to microgestures for clues about what is going on under the hatch. A twitching antenna or waving foreleg, however, will not speak for itself: any candidate sign of insight must be distinguished from ordinary exploration.
Bumblebees do not have facial expressions like ours, but Akmeşe, who has logged hundreds of hours watching them, says they are not unreadable.
“They are kind of like cats. When you spend so much time with them, you can kind of see when you annoy them” or when their mood seems to change.
The challenge is turning these impressions into movements that can be measured.
Reading the language of a bee
Loukola and his team plan to use high-resolution cameras to watch for the movements that accompany a possible breakthrough.
“They can open their mandibles, they can move their antennae and they can shake their bum or wave their forelegs,” Loukola says. “Machine-vision algorithms could pick up tiny microgestures that we cannot see with our bare eyes.”
Sensors inside the ball might also reveal whether bees push differently when rolling for play and for tool-related business. Together, these clues could help distinguish a purposeful roll from a recreational one – and bring the team closer to catching an insect in the act of figuring something out.
The researchers hope that their results will inspire respect and empathy for bumblebees, which, despite being major pollinators, are often billed below their treat-producing cousin, the honeybee.
The study cannot tell us whether a bee experiences a human-like flash of insight. But it does make the old image of insects as tiny robots more difficult to defend. A sesame-seed-sized brain may keep its secrets – but it can apparently do more than follow a script.
“Hopefully, the more you know about insects, the more willing you are to protect them,” Loukola says.
