Bee-yond numbers

On a summer evening in 1990, two research scholars sat discussing their work on bees in a pub near Wittenberg in Germany. Lars Chittka and Karl Geiger, from Freie Universität Berlin, were doing their fieldwork in eastern Germany. After a few drinks, an idea came to them that sounded like fiction. It was known that bees could learn about the colours and scents of flowers, as well as the location of their hives. But could they be trained to count? The discussion led to a new experiment, which revealed that bees, indeed, had numerical ability.

Chittka, now a behavioural ecologist at Queen Mary University of London, continues to explore apian behaviour. In a recent study (bit.ly/bees-pattern), he and his group reported that bees were good at identifying shapes and figures. Their experiment showed that bumblebees used their vision to distinguish between two similar shapes: a plus sign (+) and a multiplication sign (x), which is simply a + sign rotated by 45 degrees. The researchers trained bees to recognise and choose one pattern over the other, offering a sugar reward for the correct choice. During the experiment, the bees were allowed to fly freely in front of the patterns, while high-speed cameras recorded their movements, how long they looked at each pattern, and which parts of the patterns they focused on. After the training, the bees were also evaluated with incomplete versions of the patterns to see if they could still recognise them.

The results showed that bees didn't look at the whole pattern randomly. Instead, they focused on specific parts of the shapes — often the lower or left sections — before making a decision. The regions they inspected differed for the two signs, but each bee used a consistent scanning strategy for the same pattern. Even when the whole pattern wasn't visible, the bees could still identify the correct one by focusing on the parts they had learned before. 

The study reveals that bumblebees use an active form of vision. This active scanning helps them make quick and accurate visual decisions and evolve efficient ways to process complex visual information.