Bees help break the mystery of optical illusions

The findings, published in Proceedings of the Royal Society B, unlock better understanding of how the brain and vision systems evolve and work in humans, other vertebrates and invertebrates. They could also be used to help teach machines to judge size in complex environments. 

Led by RMIT PhD candidate Scarlett Howard, the study trained honeybees to react to a well-known visual trick called the Ebbinghaus illusion.

This trick works by making two circles the same size appear different sizes by placing smaller or larger circles around them.

“Interestingly, there is often a wide variation in the way different human subjects and some important animal models perceive illusions like the famous Ebbinghaus size trick,” Howard said.

“Besides humans, other animals that can see it include bottlenose dolphins, bower birds and domestic chicks.

"And the Himba people, a remote community in Namibia, experience a strongly reduced effect to the size illusion compared to Western and urbanised populations. 

“But up to now, we didn’t know why.”

The honeybee is an established supermodel for understanding how vision operates, especially for viewing “flower-like stimuli”.

By changing the bees’ viewing conditions for the Ebbinghaus test, the researchers found that perceiving visual illusions comes down to viewing distance.

“Some bees were allowed to fly freely when viewing the optical illusion, and subsequently perceived the illusion in a similar fashion to humans, and even dolphins!” Howard said.

“However, another group of bees that we constrained to viewing the stimuli at a set size during initial training did not see the illusion. They always judged the absolute size accurately, despite the differing contextual information.  

“The results demonstrate that visual perception is influenced by the ability of bees to choose their own viewing distance and show context is a major factor in size processing tasks.”

The results also provide insight into the evolution of visual mechanisms across vertebrate and invertebrate animals.

“Since humans see such illusions, and we are very good at complex vision (like driving a car), we suspect illusion perception to be part of higher level processing of relationships,” Howard said.

“From the current study we now know this is not something special to primate cortex; but must be wider spread in biological vision.”

RMIT’s Professor Adrian Dyer, who also worked on the study, said the results also shed light on how flower patterns influence decisions made by bee pollinators. This information could be applied to machine vision.

“Illusionary perception is important for the way in which honeybees perceive and asses the complex, dynamic environments in which they live,” Dyer said.

“It is important to understand how animals with different brains solve these problems since size processing is very problematic for current machine vision.

“We hope to learn improved ways such information can be managed and solve real-world problems for machine vision like robot guidance, process control, automatic inspection and AI facial recognition.”    

The study was led by researchers at RMIT, working collaboratively with colleagues from the University of Melbourne and the University of Toulouse in France.

The findings are published in Proceedings of the Royal Society B: http://dx.doi.org/10.1098/rspb.2017.2278  

Story: James Giggacher