When Adrian Dyer, Christa Neumeyer and Lars Chittka found that bees could recognise human faces in 2005 (J. Exp. Biol. vol. 208, pp. 4709-4714), they were truly astonished. How could such tiny brains process something as complex as a human face? And why? Was this some cute party trick, or are bees genuinely able to recognise complex images as part of their daily chores? After all,foraging honey bees have to negotiate cluttered fields and forests while searching for nectar. Can they use this remarkable visual talent to pick out and recognise one tree (as a landmark) from a forest of thousands: in other words `can they see the trees from the wood?' asks Dyer. Curious to know if bees have the ability to learn to recognise a complex tree shape, Dyer and his colleagues, Marcello Rosa and David Reser, set the bees a biologically relevant challenge: could they learn to recognise and distinguish between tree photographs (p. 1180)?
First Dyer set out to see if bees could learn to discriminate between a tree photograph they had been trained to recognise and a photograph they had been trained to avoid. Enticing a bee to land on a spoonful of sucrose, Dyer carried the insect over to a screen where he could show them two different tree pictures, each with a small landing platform for the insect to alight on. Eventually the bees learned to head to the screen on their own, and when Dyer was convinced that the bee was taking genuine interest in the images, he began training them to recognise one of the pictures by rewarding them with a nectar treat, but training them to avoid the other by presenting them with a bitter quinine drink.
After 120 training visits, Dyer took away the sucrose and quinine, forcing the insect to rely on their eyes to find the image they had been trained to recognise. Moving the images around, Dyer was delighted to see that the bees homed in on the image they had been trained to recognise. And when he presented the bees with the choice between the image they had been trained to recognise and a second novel tree picture, the insects continued returning to the tree image they recognised. Not only could the insects discriminate between tree images, but they could recognise the image from another similar picture.
So the bees seem to be capable of learning to recognise complex natural shapes. But it took an entire day to train the insects to recognise one tree image. Can they really learn to recognise complex shapes during their routine activities when they have to learn the positions of several sites during a day? This time Dyer replaced the quinine drink with water and reduced the number of training sessions from 120 to 40. Amazingly, this new group of bees correctly learned which tree to recognise, despite the milder training regime.
Having shown that bees can discriminate between, and recognise, complex natural images such as trees, Dyer is keen to see if they can do it for real in the field. But given the challenges of training and tracking insects in the natural environment, he suspects that this could take some time.