Even with a brain the size of a sesame seed, honeybees (Apis mellifera) are capable of learning something about a phenomenon that, although simple, is the foundation for humans’ ability to think and communicate through symbols. In experiments conducted by psychologists Antonio Mauricio Moreno and Deisy de Souza, of Brazil’s Federal University of São Carlos (UFSCar) in the state of São Paulo, in collaboration with entomologist Judith Reinhard, of the University of Queensland, Australia, bees interested in getting a sip of sugar water learned to make a choice based on an arbitrary relation between two types of signals (colored and striped cards)—as arbitrary a relation as that between the word ball and the actual object to which it refers.
“The research is important for learning studies because it confirms that invertebrates are capable of learning arbitrary relations,” says Dora Ventura, a psychologist and animal vision specialist at the University of São Paulo (USP), who did not participate in the experiments. Prior to the study, which was published in December 2012 in the journal PLoS One, other results had suggested that bees could also exhibit this type of learning, the simplest of what are known as pre-symbolic behaviors because they are a prerequisite for language development. But evidence of this capability needed to be obtained. The work of the group in São Carlos lends strength to the idea that even the brain of a bee—larger than that of many insects but with fewer than a million neurons (humans have 86 billion)—is capable of such a feat, long considered the exclusive domain of vertebrates with a much larger brain, such as monkeys and humans.
Honeybees began attracting the attention of researchers in the 1940s, when Austrian zoologist Karl von Frisch described a behavior unique to the species: their famous dance, a complex choreography that bees exhibit when they return to the hive and tell their companions the location of the flowers they have found. Since then, dozens of behaviors have been observed, indicating a repertory comparable to that of birds and mammals that includes the ability to count to four.
Experiments with bees differ from those with other animals. Rats and pigeons are caged and deprived of food before training. Bees are free to return to the hive, which is placed outside the laboratory. “At the beginning of the experiment, I put a feeder with syrup made of sugar water near the hive,” explains Moreno, who has worked with bees since 2002. “If it’s not raining, the bees will discover the feeder.”
In a minute, a bee gets its fill of syrup and returns to the hive to unload it. Soon, it returns to the feeder for more. Moreno then moves the feeder from the hive into the laboratory, with a few bees trailing behind. When they reach the experimental apparatus, he uses a brush with gouache ink to mark the bees he will train shortly afterward. “Sometimes they get startled and fly away, but I try to mark gently, while they are busy sipping syrup,” Moreno says.
The most elementary task that the researchers teach the bees is to distinguish between two different signals, one placed in front of a feeder containing water and another in front of a feeder with the syrup. After two hours of training, a bee learns that only one of the signals indicates the presence of sugar. That experiment, performed since the 1920s, is called simple discrimination.
The signals used are generally pairs of cards with colors, patterns or simple geometric shapes. Using as a basis studies by Dora Ventura on the visual capacity of bees, after several attempts, Moreno concluded that these insects do better distinguishing yellow cards from blue ones and black-and-white horizontally striped cards from vertically striped ones.
That does not mean, however, that bees are incapable of recognizing more complex signals. In 2010-2011, while Moreno was doing part of his doctoral research at the University of Queensland, he collaborated on another study conducted at Reinhard’s laboratory, in which Apis mellifera learned to differentiate paintings by Picasso from those of Monet. In simple-discrimination experiments using cards with reproductions of various works, one group of bees got syrup if they chose the Cubist over the Impressionist, while the other group was trained to make the opposite choice, and both groups were successful.
To a bee, anything is worth getting to the syrup. If, for example, the syrup is always on the left, the bee may respond correctly because it memorized its position, and not because it distinguished the visual signals. To avoid that possibility, the position of the feeders was changed constantly. The feeders were also frequently replaced by new ones so the bees would not be guided by the scent of the drops of syrup.
During the next phase, Moreno and his colleagues conducted what are known as conditional-discrimination tests, in which they added a step to the experiment. Before coming across the blue and yellow signals, for example, the bee encountered a striped card in front of it. When the stripes were vertical, the syrup was behind the yellow card. If the stripes were horizontal, the correct card was the blue one. For the bee to do well, it needed to deduce that arbitrary relation—in other words, to understand that the color choice depended on the condition of the stripes.
After six consecutive days of training, the honeybees were able to correctly make about 70% of the conditional discriminations. Even when the feeders were removed, they chose correctly.
Encouraged by the results, the researchers decided to see whether the bees could go farther and learn another prerequisite for symbolic thought: creating new relations based on previously-learned arbitrary relations.
They then modified the test by reversing the order of the cards. They wanted to see whether, for example, a bee that was trained to choose yellow if it had previously passed through vertical stripes was capable of choosing the vertical stripes when passing through a yellow card. The answer was no. “On this point I’m not optimistic,” Moreno says. “I don’t know how we could produce more complex behaviors than what we obtained.”
Back in Brazil, Moreno conducted more conditional-discrimination experiments in de Souza’s laboratory at UFSCar, this time testing Brazilian stingless bees (Melipona rufiventris), which enabled them to forgo the use of the beekeeper’s suit needed for Apis mellifera.
Although they discriminated the cards well, nearly all the stingless bees that were tested failed to learn the arbitrary relations between them. “In a similar experiment conducted a year earlier, just one bee learned, but only after three weeks of training—its entire life span,” Moreno recalls. “It died soon afterwards.”
That result contradicts the pioneering animal behavior experiments conducted by psychologist and novelist Isaias Pessotti in the 1960s at USP in Ribeirão Preto. Pessotti, who invented an automatic device that signaled the bees using colored lights next to feeders opened and closed by small levers pressed by the insects themselves, concluded that stingless bees were capable of conditional discrimination. In Pessotti’s device, however, the conditioning and choice signals were shown simultaneously. For this reason, many researchers wonder if the stingless bees had learned to choose pairs of signals as if they were a single thing, rather than establishing a conditional relation. To eliminate that possibility, Moreno and his colleagues presented the colored and striped cards one after the other rather than at the same time.
Moreno, de Souza and Reinhard have ventured to provide an explanation for the apparent superiority of the honeybee. Because Apis mellifera is native to a temperate climate, it may have evolved to be capable of making more-complex associations, such as between the seasons of the year or the flowerage of different plant species. Stingless bees live in smaller colonies that do not need a wide variety of flowers for survival. Moreover, tropical flowerage is more constant throughout the year, so there would be no reason for stingless bees to vary their choices. “But it’s only speculation,” Moreno points out. “We need more studies comparing the foraging habits of Melipona and Apis.”
“The negative outcome with Melipona shouldn’t be taken too seriously,” says Randolf Menzel, a neurobiologist and bee specialist at the Free University of Berlin, Germany. As the authors of the paper acknowledge, Melipona’s lack of success in the experiments could have been caused by unknown factors.
“The experimental apparatus could have been perceived differently by the two species,” explains Martin Giurfa of the University of Toulouse, France, another authority on bee behavior and neurophysiology. “The experiment could have been more stressful for the Brazilian bees and therefore diminished their performance.”
1. Matching with a model in bees (Melipona quadrifasciata) (nº 2008/50576-8); Grant mechanism Doctoral research grant; Coord. Antonio Mauricio Moreno; Investment R$ 132,486.12 (FAPESP).
2. Institute for the Study of Behavior, Cognition and Teaching (nº 008/57705-8); Grant mechanism Thematic project; Coord. Deisy das Graças de Souza (UFSCar); Investment R$575,983.91 (FAPESP).
MORENO, A.M. et al. A comparative study of relational learning capacity in honeybees (Apis mellifera) and stingless bees (Melipona rufiventris). PLoS One. v. 7 (12). 2012.